bipolar power supply IC +/-5v 750ma

[benbiles]

HI, I'm looking to make a power supply that can run 8 PGA2500 audio preamps. They run on +/- 5V and they
use up too 85ma each so I'm looking to make 680ma ( 700ma ) with some headroom from around 9-16.5vdc.. ( 12v litthium power )

I found some 100ma ( 50ma per + / - ) bipolar ic's form LTC. I could use 1 per IC or run them in parralel ? but this seams like a waste of PCB spce.

Can any one sagest a good way to do this ? I was thinking I should use a sepperate LDO's, one for + and one for - power. are there any downsides pitfalls to this ? if not any idea which IC's would be a good candidate?

The power needs to be low noise since I'm building a low noise muli-channel pre amp.

Thanks,

Ben

 

[KlausST]

Hi,

It seems you want to build some sophisticated audio equipment.
If all channels are on one PCB, you don't have to take care about ground loops (between amplifiers, signal lines and power supply),
therefore i'd keep it simple and build a common supply.

To avoid crosstalk via the power lines you could use individual - for each supply rail and each amplifier - RC low pass filters (or damped LC filers) .

For a battery powered device I recommend to use switching supplies. Use high quality controller ICs for stable and noise free supplies.
Often it is recommended to use LDOs after the switching regulators. With a good switching design, I think it is not necessary.

More critical is parts placement, PCB layout, power path routing, ground plane splitting.

Good luck

Klaus

 

[benbiles]

hi, any idea which common supply IC can do the job? couldn't find anything that could deliver more than 100ma +/-5v and need 750ma!


The 8 channels are on the same board, digital power comes from another board and digital ground plane will be split if not on another board entirely !

Hi,

It seems you want to build some sophisticated audio equipment.
If all channels are on one PCB, you don't have to take care about ground loops (between amplifiers, signal lines and power supply),
therefore i'd keep it simple and build a common supply.

To avoid crosstalk via the power lines you could use individual - for each supply rail and each amplifier - RC low pass filters (or damped LC filers) .

For a battery powered device I recommend to use switching supplies. Use high quality controller ICs for stable and noise free supplies.
Often it is recommended to use LDOs after the switching regulators. With a good switching design, I think it is not necessary.

More critical is parts placement, PCB layout, power path routing, ground plane splitting.

Good luck

Klaus

 

[KlausST]

Hi,

You are kidding...

Go to any manufacturer's or distributor's site and find hundreds.

Klaus

 

[benbiles]

more info.. sorry , not kidding.. ( beginner alert ! )

input vdc 16.5v - 10.5v lithium battery
output needed: dual rail 5v +/-
850ma - 1amp power
Low noise ripple for audio preamps

search digikey.com etc and did'nt come up with anything in that input voltage range and
output power in both LDO and switched power supply IC catagorys.

Think I must be doing something wrong here? should I be searching for two IC's , one for positive and one for negative supply rather than dual raiil?

Hi,

You are kidding...

Go to any manufacturer's or distributor's site and find hundreds.

Klaus

 

[D.A.(Tony)Stewart]

What are your ripple specs for +/-5V @9V with +9.5V input?

Budget? Minimum Efficiency?

Is this the best battery you can select with no centre tap?

 

[benbiles]

quiet as possible .. 10 - 20 mv ?

cost , not so important
weight, very important ! no heavy transformers..
-40c to 85c ( or near )

its supposed to be a portable proffessional audio device. power efficiancy important. should be low noise through vdc input range of 14.5v - 8.5v ( total possible range 16.5v - 8.5v )

Power is for 8 x PGA2500 preamp / gain controller ic's, 85ma per preamp but think I need headroom.

searching for bipolar Ic's in this catagory but don't see many options.

Is it bad to use one ic +5v 500ma and one -5v 500ma off the same Vdc supply? I'm thinking that could cause more noise ( 2 x ripple unless in sync ? )

or maybe it smooths out the ripple ??

no choice on battery spec / its a standard pro audio battery that has roughly this spec but they specs vary a lot..

Chemistry Lithium-Ion
Output Voltage 14.6V
Connection NP-1 Type Battery
Capacity 68Wh
4.6Ah

What are your ripple specs for +/-5V @9V with +9.5V input?

Budget?

 

[D.A.(Tony)Stewart]

what about efficiency for battery operation?

With 9.5V in and 10V (+/-5) out increases cost, complexity


What about +/-7.4V input +/-1V range for LiPo
Does charger have charger voltage balancing?
Is load even or asymmetric +/-


Is this a MAKE (design) or a BUY

This topology would support your requirements. with <20mV ripple 1~2% balanced with effort....
Consider buying from Murata. Call them.

$27.57USD (100pc) https://ca.mouser.com/ProductDetail...C2S03fHO9fYh3Dv%2bDgIn5l%2b5QlOfI/VQdDW4/EA==

 

[KlausST]

Hi,

... you were looking for one part generating both supply rails. These are unusual, but there are some.
If you want to look at farnell:
Home > Semiconductors - ICs > Power Management > DC / DC Switching Regulators - Multi Output

****

Imagine: there are two rails. both rails need to be regulated, a single regulation loop won´t give satisfying results, because the load on both rails is not equal.
So if the load is independent you need the regulation loops to be independent, too.

****

Or you could look for a module like: TRACO TEN 8-1221

Klaus

 

[Audioguru]

You are mistaken. A NP-1 Type Battery made for cameras has 4 Li-Ion cells. Each cell is 3.2V when dead or 4.2V when fully charged. Then it will have 12.8V to 16.8V. How will you divide its output so that you can have plus and minus 5V?

 

[benbiles]

Hi Audioguru, yes your correct, the NP1 was origianlly made for sony video camera's! It is now used and has been used for many years along with other battery designs to power location mixer recorders. yes some of the NP1 designs are 16.8v - 12.8v and some are lower. some IDX versions of NP1 is 14.8 - down.. It has in built cicuit overload and most likely regulator of some kind. There are Bipolar switching dc-dc regulators that can achieve the +/-5v as mentioned..

SunnySkyGuy , The charger unit is not part of the design, off mixer.

I am building a prototype and drawing up / changing the pcb design in altium designer as I go along. I am hoping the design will work in modular form. Preamp ( gain control ) - DAC ( 8 channel x 2 ) - 8 x 2 mix ( master volume / headphone volume ) DAC section is done and working nicely on 5v / 2.5v dc on split ground plane. Thanks for the murata link.. looks good !!

KlausT, so I can just go for 2 x low noise LDO's 500 ma each based on that logic ! ? the link also looks good although quite narrow margin for overhead.. will take a better look at PGA2500 data sheets, mayeb I dont need the overhead... 88% efficiancy looks great..

will have a good look through Farnell tmrw.. 1:53am here in Japan !

 

[Audioguru]

14.8V is 3.7V per cell of a 4-cells Li-Ion battery where it is half-discharged. Many lithium batteries have a protection circuit with them to prevent over charge, over discharge and over temperature to prevent an explosion if something goes wrong. The protection circuit is not a charger circuit.

 

[KlausST]

Hi,

KlausT, so I can just go for 2 x low noise LDO's 500 ma each based on that logic ! ? the link also looks good although quite narrow margin for overhead.. will take a better look at PGA2500 data sheets, mayeb I dont need the overhead... 88% efficiancy looks great..

Yes, a generator for center voltage (gnd) and two LDOs can work. When the LDOs have low gnd current you may achieve about 70% efficiency.

Klaus

 

[BradtheRad]

Here is a way to split a single supply into 2 bipolar supplies. Normally it would be a voltage doubler, however you can adjust values to obtain a lesser output voltage.



The clock pulses are over-simplified. You'll probably want to add control circuitry, to alter frequency, duty cycle, etc., in order to maintain output at 5V to each load. This will keep voltages balanced when your loads are unequal.

- - - Updated - - -

The inductor is not absolutely necessary, but without it there are large current spikes as capacitors charge quickly.

With the inductor you have a buck converter, or rather, dual buck converters with alternating action.

 

[benbiles]

Hi Klaus, going back to one of your erlier comment,

To avoid crosstalk via the power lines you could use individual - for each supply rail and each amplifier - RC low pass filters (or damped LC filers)..

the IC's can have RC filter on power inputs.. but can you explain what crosstalk is ? is that caused by mismatch of loads and differnt supply line resistance or something? If this effects noise levels then maybe I should just have one power IC per PGA ? theres 7 caps and 4 resistors per supply so I suppose it does'nt take up to much realestate in SMD!! is there a disadvantage of doing that regarding efficiancy? I can get 100ma per LTC3260 ic. would post a link to the datasheet but not sure that ok on this forum..


BradtheRad, not sure how that works, can you explain how this works? voltage dividers and diode ? mmm confused.. and how do I crate the sqare wave input? think that might be out of my depth..

Here is a way to split a single supply into 2 bipolar supplies. Normally it would be a voltage doubler, however you can adjust values to obtain a lesser output voltage.



The clock pulses are over-simplified. You'll probably want to add control circuitry, to alter frequency, duty cycle, etc., in order to maintain output at 5V to each load. This will keep voltages balanced when your loads are unequal.

- - - Updated - - -

The inductor is not absolutely necessary, but without it there are large current spikes as capacitors charge quickly.

With the inductor you have a buck converter, or rather, dual buck converters with alternating action.

 

[KlausST]

Hi,

.. but can you explain what crosstalk is ?

Crosstalk is when the signal from one channel influences another channel. The influence usually is a very low level.

With single ended signal path this could be that a signal at one Opamp causes varying current in the supply lines. The current will lead to a varying voltage in the supply lines. If this is a common supply line for another channel this may cause an influence in it's output signal (PSRR).

With your differential signal processing it is very unlikely that there is an audible influence caused by common supply rails.

Klaus

 

[BradtheRad]

BradtheRad, not sure how that works, can you explain how this works? voltage dividers and diode ? mmm confused.. and how do I crate the sqare wave input? think that might be out of my depth..

The transistors turn on alternately, charging the capacitors alternately. Each capacitor receives all (or almost) of the supply voltage. Your new ground is at the node between the capacitors.

The familiar 555 timer IC can provide a pulse train. My schematic shows a voltage being applied to the 'ctl' pin. It is a handy way to adjust the duty cycle coming from the 555. You may want 40% duty cycle, or 50, or 60.

The whole concept will require a lot of experimentation, fiddling with this or that component value, or frequency, etc. Additionally since your power supply is variable, you'll need to add the necessary regulating circuit. It will be a challenge. It should turn on the transistors at the proper frequency and duty cycle, so that you get smooth 5V at each load, under all conditions.

Anyway after considering all that you may decide it's easier to go with another method suggested in someone else's post.

And there are power converter modules available that will make the job easy.

 

[benbiles]

Bradthrad: thanks for explination, that makes sence, did'nt get the idea of the square wave pulsing the transitors on/off alternately before, clever stuff !

I'll will try and go for a one IC solution and hope threres no cross talk ( or just use one LT3260 per channel )
I dont even have an osciloscope yet, so designing my own DC-DC switching power supply would be next to impossible and just a lot worse than any IC version I think !

KlauST : I tried to search farnel but I don't get any results for [ voltage output1=+5v ] [ voltage output2=-5v ] supplies under multichannel with DCV voltage input low high. I tried both UK and USA websites, then even tried using a VPN but they block that entirely.. its probebly a security thing..

I'll just have to go through the manufacturors websites..Emailed Murata to see what they say.

will peobebly just go with the TRACO TEN 8-1221 , that would power 8 x pg2500's and its all in a small shielded box.

now i just need to work out if I can digitally pan 8 channels from the DAC without doubling up the 8 chanel DAC think thats another topic though !!

The transistors turn on alternately, charging the capacitors alternately. Each capacitor receives all (or almost) of the supply voltage. Your new ground is at the node between the capacitors.

The familiar 555 timer IC can provide a pulse train. My schematic shows a voltage being applied to the 'ctl' pin. It is a handy way to adjust the duty cycle coming from the 555. You may want 40% duty cycle, or 50, or 60.

The whole concept will require a lot of experimentation, fiddling with this or that component value, or frequency, etc. Additionally since your power supply is variable, you'll need to add the necessary regulating circuit. It will be a challenge. It should turn on the transistors at the proper frequency and duty cycle, so that you get smooth 5V at each load, under all conditions.

Anyway after considering all that you may decide it's easier to go with another method suggested in someone else's post.

And there are power converter modules available that will make the job easy.