[joneeboi]

What about the TI DCV010505D (datasheet: http://focus.ti.com/lit/ds/symlink/dcv010505.pdf)? It gives a +/- 5V supply with 100 mA on each rail. That sets ground at USB ground which allows us to safely connect the DAC, amp and case together as the ground, gives us more current to work with, and is more efficient than using one rail on a dual rail chip (right?). Am I reading that right? Is it 100mA on each rail or for both? I guess another drawback is its 1W rating and low stock at Mouser and Digikey.

I'm wondering, if we set a virtual ground at 6V, will the volume knob short against the case?

 

[rds]

Quote:

200ma is not sufficient for this?

At this point in time charge pumps are used almost exclusively for digital supplies where the various non linearities, noise, and v-droop are not an issue.
If you want to use one, no one is stopping you. But consider that circuit designers won't use these in even the lowest quality of analog circuits.
Besides that, the your bipolar supply would be totally asymmetrical.

 

[rds]

Quote:

What about the TI DCV010505D

It should be able to provide 100mA on each rail max. But compare the performance curves of the dcp and dcv. The output ripple for dcp at the 140mA max we'll be using is about 20mV with a 1uF cermanic. Now for the dcv using 70mA per rail there's about 35mV ripple with a 1uF cap. The Vout also drops much more steeply in the operating range we'd be using for the dcv than the dcp. So you really need regulation and the most you could regulated is +/- 3.3V. So you see that in the end you don't gain a lot of voltage.

It's also worth mentioning that 35mV ripple on each 5 V rail is far more significant than 20mV ripple on a single 15V rail. Think of the ripple to DC ratio.

The reason I would prefer a single supply is that most of the amps we'd be using this with take a single supply.

Quote:

That sets ground at USB ground which allows us to safely connect the DAC, amp and case together as the ground,

I think you're getting too caught up with the idea of 'ground'. The ground out of the converter is not 'at usb ground' it is totally isolated. It is at whatever potential you connect it to, and the rails are just +/-5V of that potential.

 

[error401]

Quote:

Originally Posted by rds /img/forum/go_quote.gif No I don't plan +/- 12V. There's no point in trying to get that kind of voltage off a usb. The current would be useless as you pointed out. I'm using one output of the DCP020515DP. I want a single 12V supply. That means we can draw 148 mA at 15V and 89% efficiency. The ldo is rated at 150mA as well. I estimated the upper limit and 150mA for this reason. Maybe I'm pushing it. But for sure 140mA is reasonable. It does require the builder to think about current draw if they're getting close to the 140mA mark, but as an example you could even build a 12V PPAv2 with the buffers biased at 25mA (near the max) and you'd be OK. So it will be pretty robust in terms of powering headphone amps.

My mistake, I apologize. I'm not sure you can just ignore the second output and expect to get 2W out of one half of it.

 

[joneeboi]

The TI notes in the datasheet say you can hook them up in parallel to double current capability.

 

[error401]

Quote:

Originally Posted by joneeboi /img/forum/go_quote.gif The TI notes in the datasheet say you can hook them up in parallel to double current capability.

I'm not sure that will work with the dual chips. There's only one transformer in the dual chips, with a modified rectifier configuration. I think you'd end up with a shorted secondary.

 

[rds]

No it will not work, becuase the output ground is common.

But it looks like you're right about about the current limit. We can only draw a max of 66mA per output on the +/-15V chip.

It seems the only way to get a decent current and voltage is to use two DCV010515. That will give 134mA into 15V. However, it's a little cost prohibitive at $20 for the 2 chips.

EDIT I will stubbornly go ahead with the +/-15V chip I already ordered and make a +/-12V supply that can supply 37mA per channel at usb 2.0 specs. That'll be cutting it really close for a pimeta with lmh6321. But we'll see what happens.

The 2 chip supply mentioned above is a good alternative to my original 15V idea, and I will try that at some point.

 

[joneeboi]

In the spirit of redesigning my own HPDAC-type board, why not use the same DC booster the HPDAC did, the TPS61040? The parts count would be a bit higher, but it comes in 250mA and 400mA packages while showing great stock at Digikey and Mouser. It's configurable up to 28V, so we can set it just above whatever filtered and regulated voltage we wanted. I mean, all the info is on dsavitsk's site for us to use...

 

[rds]

The HPDAC looks like a nice project. If you want to duplicate it why not just ask dsavitsk for the pcb files?

I'd like to try something different and my plan is still to make a usb pimeta with bantam dac.
I will probably end making the two chip supply at some point, and swapping out the +/- 12V, unless of course it works perfectly.
Even still I will probably still try the 12V 130mA supply - which would be nice because it's not so application specific.

 

[millwood]

Quote:

Also consider that the really crappy power only applies to one rail.

you can easily create dual rails (or quad rails if you wish) from a charge pump.

they just don't hold their rails that well under a dynamic load.

 

[millwood]

Quote:

What about the TI DCV010505D (datasheet: http://focus.ti.com/lit/ds/symlink/dcv010505.pdf)?

TI offers three families of those tiny dc-dc converters.

the first one is the dcp01/02, using bipolar processes. there is the regulated version in dcr (3.3v/5v only), and then the new dcv (bicmos) for higher efficiency.

Quote:

is more efficient than using one rail on a dual rail chip (right?).

you are right on that one.

Quote:

But compare the performance curves of the dcp and dcv.

if you look at the ripple charts later in the datasheet, they have fairly comparable ripple performance and regulation as well.

Quote:

I think you'd end up with a shorted secondary.

the difference between the dual and single Vout versions is just full wave rectification vs. half wave rectification.

Quote:

TPS61040

that is non-isolated so if you mess up on the boost side of the circuitry, you could get your computer / usb ports damaged.

ON has a few nice parts like the TPS61040 in sot23-5 packaging - it helps greatly with board real estate.

 

[rds]

Quote:

TI offers three families of those tiny dc-dc converters. the first one is the dcp01/02, using bipolar processes. there is the regulated version in dcr (3.3v/5v only), and then the new dcv (bicmos) for higher efficiency.

You should make sure you read things carefully before you start giving advice.
All 3 are cmos/dmos. The dcv is rated to higher isolation voltage. The dcp02 is higher wattage and has better load regulation. The dcv is not the new dcp02. In fact, there's no reason for us to use it as we don't need huge isolation voltage.

The better load regulation of the dcp02 is why the performance curves look better - they are better.

I don't mean to sound like a jerk, but it is kind of annoying when people jump in and start saying this stuff in very matter of fact way when they don't have their facts straight.

 

[error401]

Quote:

Originally Posted by millwood /img/forum/go_quote.gif the difference between the dual and single Vout versions is just full wave rectification vs. half wave rectification.

Yea. And when you connect the two half-wave rectified outputs together, you short the secondary through the two diodes...

It's unfortunate there isn't a more appropriate voltage output for them, I remember having the same problem picking one when I did it. Even a single-output at 15V, 2W would be much better than this.

Quote:

why not use the same DC booster the HPDAC did, the TPS61040?

Virtual ground.

 

[millwood]

Quote:

Originally Posted by rds /img/forum/go_quote.gif You should make sure you read things carefully before you start giving advice.

we all could use some of that, right?

Quote:

All 3 are cmos/dmos.

I think TI would beg to differ. here is exactly what they said:

Quote:

Originally Posted by TI AN: sbva013a.pdf The DCP01B, DCV01, and DCP02 are three families of miniature DC/DC converters providing an isolated unregulated voltage output. All are fabricated using a CMOS/DMOS process with the DCP01B replacing the familiar DCP01 family that was fabricated from a bipolar process. The DCP02 is essentially an extension of the DCP01B family providing a higher power output with a significantly improved load regulation, and the DCV01 is tested to a higher isolation voltage.

so is that enough to establish the fact that dcp01 (no B) is a bipolar process?

As to dcp02/dcv01, well, here comes from their datasheet:

Quote:

Originally Posted by TI datasheet for DCV The circuit design utilizes an advanced BiCMOS/DMOS process.

Quote:

Originally Posted by TI datasheet for DCP02 The circuit design uses SYNC pins together, taking care to minimize the an advanced BiCMOS/DMOS process.

I would take a wild guess that Bi in BiCMOS means Bipolar?

Quote:

But compare the performance curves of the dcp and dcv. The output ripple for dcp at the 140mA max we'll be using is about 20mV with a 1uF cermanic. Now for the dcv using 70mA per rail there's about 35mV ripple with a 1uF cap.

here is a comparison of two particular chips, dcp020505 vs. dcv010505, straight out of TI's datasheets for those respective chips.

at 50% load, for 1uf cap, dcp02 has a ripple of <25mv, and dcv01 has a ripple of ~18mv (my estimate). so ~18mv is quite comparable to <25mv, right?

at 100% load, for 1uf cap, dcp02 has a ripple of <50mv and dcv01 has a ripple of ~50mv. so ~50mv is quite comparable to <50mv, right?

so how do we recouncile your numbers to TI's?

Quote:

The dcp02 ... has better load regulation.

let's talk about regulation.

look at the 2nd attachment. from 20% load to 100% load, dcp02 has a regulation of 6% (106% to 100%), and dcv01 has a regulation of 8% (108% to 100%). better? yes. material? probably not.

and if you look a few pages earlier, both chips are quoted to have 1% regulation (constant Io).


Quote:

The dcv is not the new dcp02.

nobody said it is. better reading would have helped here.

Quote:

In fact, there's no reason for us to use it as we don't need huge isolation voltage.

nor the higher wattage? nor the 2% better regulation? ...

Quote:

I don't mean to sound like a jerk, but it is kind of annoying when people jump in and start saying this stuff in very matter of fact way when they don't have their facts straight.

agreed. having your facts straight does help.

 

[millwood]

Quote:

Originally Posted by error401 /img/forum/go_quote.gif Yea. And when you connect the two half-wave rectified outputs together, you short the secondary through the two diodes...

I have a very difficult time understanding how that could be the case.

could you elaborate please?