Heatsink-less opamp power supply

I've been working on trying to make a simple opamp power supply that has high efficiency - high enough to go without a heatsink.

I am basing it around the TPS7A* regulators from TI and the Bourns CD-HD2004 schottky diode bridge.
The maximum dropout of this pair should be ~1.3V at the max load of 1A (typical is ~0.815V). Reducing the load to 500mA should reduce the dropout to ~0.78V

So on to the transformer selection. Ideally, I dont want to have all these efficient parts only to have to deal with a load of excess from the transformer.
12VAC is a common tx value and would give me an RMS of 17V. I think this may be cutting it too fine for a +/- 16V output.
So, I thought; how about a transformer rated at 12V AC for a larger-than-needed current? (e.g. 12VAC @ 2A), this would, in theory, give me a slightly higher VAC at lower current draw. Is it unreasonable to rely on 'side-effects'?

schematic:
https://ramshackleaudio.s3.amazonaws.com/media/images/ps_1.original.png

https://ramshackleaudio.s3.amazonaws.com/media/images/ramshackle_power_supply_1.original.png

Have you factored in ripple voltage on the reservoir caps ?

JR

Have you factored in mains voltage fluctuations?

Cheers

Ian

The most efficient power supply is one without regulators. Depending on the application, it may be perfectly suitable. I've been guilty of producing many level translators and even phono preamps that used unregulated supplies. Your 12Vac xfmr is perfect for delivering +/-16V nominal, +/-18 on a sunny day and +/-14 on a rainy day.

What is your objection to heatsinks??

How MANY opamps do you have, to need 0.8 Amps?

ramshackles said:

I've been working on trying to make a simple opamp power supply that has high efficiency - high enough to go without a heatsink.

I am basing it around the TPS7A* regulators from TI and the Bourns CD-HD2004 schottky diode bridge.
The maximum dropout of this pair should be ~1.3V at the max load of 1A (typical is ~0.815V). Reducing the load to 500mA should reduce the dropout to ~0.78V

Click to expand...

Those regulators do need heatsinks! But instead of the usual aluminum things to which you bolt a TO-220, you implement the heat sink with copper pours on the PCB.

I have used those regulator but as post regulation of a switching converter with less than 500mV dropout.
With a traditional transformer, bridge, cap PSU, you have to deal with mains variations, transformer load factor, ripple voltage ... as Ian, JR and Abbey said.
It is then very difficult to design with less than 2V dropout in the worst case which means 2W power dissipation for 1A. It is possible for a TPS7Axxxx but with a 4 layer PCB with a solid ground plane.

Opamps does not really need regulated power supplies in most case, good filtering is often enough, you can also try cap multipliers, or more complicated designs (Kmultiplier for high loads ...). In most audio designs regulators are in fact used as filtering devices since there is no real need for precise voltage rails.

ruffrecords said:

Have you factored in mains voltage fluctuations?

Cheers

Ian

Click to expand...

Yeah, so in the UK, mains voltage can be 216.2..253V and be within spec (I think it is 230V -6 to +10%). This is the main reason why I'd be uneasy about selecting a 12VAC transformer (and also leaving no room for heatsinks at all - atm they could be mounted on the reverse of the board)

PRR said:

What is your objection to heatsinks??

How MANY opamps do you have, to need 0.8 Amps?

Click to expand...

No real objection, although it is nice to cut down on bulk & cost (a little).  Really, I like the idea of 'easy' efficiency gains offered by these chips.

You are right - 0.8 amps will power a lot of amps. So I want to measure (if I can), how much the dropped voltage reduces when running at lower than the rated loads. This could be useful for specifying at what loads a heatsink is not needed and/or you can get away with a lower voltage transformer.

Andy Peters said:

Those regulators do need heatsinks! But instead of the usual aluminum things to which you bolt a TO-220, you implement the heat sink with copper pours on the PCB.

Click to expand...

Yes..to be clear its the aluminium things I was on about. Its not clear from the PCB shot, but the footprint for those regulators has a load of through holes in the center which you use to connect the copper pour (and external heatsink if necessary/desired)

Chris_V said:

I have used those regulator but as post regulation of a switching converter with less than 500mV dropout.
With a traditional transformer, bridge, cap PSU, you have to deal with mains variations, transformer load factor, ripple voltage ... as Ian, JR and Abbey said.
It is then very difficult to design with less than 2V dropout in the worst case which means 2W power dissipation for 1A. It is possible for a TPS7Axxxx but with a 4 layer PCB with a solid ground plane.

Opamps does not really need regulated power supplies in most case, good filtering is often enough, you can also try cap multipliers, or more complicated designs (Kmultiplier for high loads ...). In most audio designs regulators are in fact used as filtering devices since there is no real need for precise voltage rails.

Click to expand...

So, I'm working up to basically using them as post regulation for switchers. I intended to explore, build, learn, mess-up etc a more simple design (while still using the TPS7A...) before eventually moving on to the more complex switching design.

ramshackles said:

Yeah, so in the UK, mains voltage can be 216.2..253V and be within spec (I think it is 230V -6 to +10%). This is the main reason why I'd be uneasy about selecting a 12VAC transformer (and also leaving no room for heatsinks at all - atm they could be mounted on the reverse of the board)

Click to expand...

As Chris_V mentioned, these VLDO regs are designed to be used as post regulators, not as the primary regulators. If you want high efficiency in a primary regulator, you need to look at switching types.
You may want to look there
https://ac-dc.power.com/products/topswitch-family/

abbey road d enfer said:

As Chris_V mentioned, these VLDO regs are designed to be used as post regulators, not as the primary regulators. If you want high efficiency in a primary regulator, you need to look at switching types.
You may want to look there
https://ac-dc.power.com/products/topswitch-family/

Click to expand...

Yup...baby steps. Im working up to that .
Anyway...since I have a bunch of transformers and reservoir caps, Im gonna pursue the simple design for my own 'enjoyment', learn and move on to the switching design.

ramshackles said:

Yeah, so in the UK, mains voltage can be 216.2..253V and be within spec (I think it is 230V -6 to +10%). This is the main reason why I'd be uneasy about selecting a 12VAC transformer (and also leaving no room for heatsinks at all - atm they could be mounted on the reverse of the board)

Click to expand...

It's been decades since I've been dealing with mains voltage variations but for the entire world it was more like +/-20%, and we would still encounter even worse outliers.

I suspect inside UK proper the mains voltage will be well better behaved than the Australian outback. 

JR

Andy Peters said:

Those regulators do need heatsinks! But instead of the usual aluminum things to which you bolt a TO-220, you implement the heat sink with copper pours on the PCB.

Click to expand...

If you're simply trying to dodge having to add parts, this is the way to go. The board can sink some of the heat or you can attach the regulator to the side of a chassis.