How to share a transformer

[Doodski]

How about adding a large capacitor (6-8000 uF) after the regulator? Could that buffer those relatively short high current spikes?

You mean six, 8000 microFarad caps after the regulator? That's massive smoothing. Not necessary. I'm not sure what you need/want for smoothing after the linear regulator but I do know that I've made linear regulated power supplies for Sony portable mini-disc players and they worked just fine without smoothing after the linear regulator. There was zero availability of the DC power accessory so customers where asking me to make them for them. I took 12VDC from a cigarette lighter plug and regulated to whatever was the requirement of the time. A car is a pretty nasty enviroment electrically and they worked just fine without post regulator smoothing caps.

 

[Doodski]

O' I just had a thought. If you check the components/devices that you want to power with this power supply that you are making I think you will find that they have smoothing/filtering capacitance at their circuitry already. The caps will be placed near the op amps and digital IC's. So adding more caps might be futile and unnecessary.

 

[dennnic]

You mean six, 8000 microFarad caps after the regulator? That's massive smoothing. Not necessary. I'm not sure what you need/want for smoothing after the linear regulator but I do know that I've made linear regulated power supplies for Sony portable mini-disc players and they worked just fine without smoothing after the linear regulator. There was zero availability of the DC power accessory so customers where asking me to make them for them. I took 12VDC from a cigarette lighter plug and regulated to whatever was the requirement of the time. A car is a pretty nasty enviroment electrically and they worked just fine without post regulator smoothing caps.

No, I meant 6000 to 8000 uF total. The thought behind was that load might draw more from the capacitors and stress (not melt) regulators less during current peaks.

 

[antcollinet]

I think for the 1.2A raspberry pi supply you might be better off looking for buck regulators -these are switching devices rather than linear and dissipate much less heat for the same power output. (Like class D amp instead of Class A)
EG:

Youmile 6PACK Step-Down Power Supply Module LM2596 LM2596S Buck Converter DC to DC 3.0-40V to 1.5-35V High Efficiency Voltage Regulator Module : Amazon.co.uk: Business, Industry & Science

Youmile 6PACK Step-Down Power Supply Module LM2596 LM2596S Buck Converter DC to DC 3.0-40V to 1.5-35V High Efficiency Voltage Regulator Module : Amazon.co.uk: Business, Industry & Science

www.amazon.co.uk

Otherwise you've got at least the 7W or so I mentioned previously in the regulator. Pre or post capacitance won't help since the current still has to go through the regulator.

On the other hand you can get a reasonably sized heatsinks. Here is one with 5.6C/W - which should limit temperature rise to around 42c above ambient - still quite toasty though.

Spiratronics Vertical Mount TO220 Heat Sink 5.6C/W : Amazon.co.uk: Computers & Accessories

Spiratronics Vertical Mount TO220 Heat Sink 5.6C/W : Amazon.co.uk: Computers & Accessories

www.amazon.co.uk

PS - are you sure 1.2A is enough? the official Pi 4 PSU is 3A

 

[antcollinet]

No, I meant 6000 to 8000 uF total. The thought behind was that load might draw more from the capacitors and stress (not melt) regulators less during current peaks.

You don't need significant smoothing caps on the output, that is what the regulator is for - and too much can destabilise the regulator. A few 10's of uF is more than adequate - close to the load.

They don't help with load on the regulator at all.

 

[antcollinet]

Here is a fixed 5V buck converter, with USB connector:

ARCELI DC 6V 9V 12V 24V to DC 5V 5A Buck Converter Module, 9-36V Step Down to USB 5V Transformer Dual Output Voltage Regulator Board: Amazon.co.uk: Electronics & Photo

Free delivery and returns on eligible orders. Buy ARCELI DC 6V 9V 12V 24V to DC 5V 5A Buck Converter Module, 9-36V Step Down to USB 5V Transformer Dual Output Voltage Regulator Board at Amazon UK.

www.amazon.co.uk

Or you could just get the official PSU for about $10

 

[Doodski]

No, I meant 6000 to 8000 uF total. The thought behind was that load might draw more from the capacitors and stress (not melt) regulators less during current peaks.

The discharge and charge rate of those caps that you suggest will be affected by the time constant(s) of the RC circuit(s) that is made. I'm not aware if using oversize caps is a design feature or not that is used for reducing the peak load on the linear regulator and even on the transformer as you mentioned previously. That is a question best answered by an electrical or electronics engineer.

 

[Deleted member 46664]

Hi,

I'm looking for the best way on how to get multiple outputs from a single secondary coil transformer.

I've got a 9 volts 50 VA secondary and would like to build several independent outputs for raspberry and it's hats.

Would there be any downsides on sharing a single secondary?
As far I as see it, I can just use regulated output and perhaps connect multiple things in parallel; use a DC output from diodes and regulate each output independently; or use AC and independent rectifiers, caps, regulation for each one... How about a second approach?

Thanks in advance,

Regards,

Stefan

Can I make an alternative suggestion here...
I would use an SMPS brick power supply for this. Start with a 24 volt 2amp supply ... add regulators and additional filtering as needed.

You mentioned needing opamp supplies for a cluster of 7 opamps ... you likely only need about 100ma for that entire circuit. Opamps draw next to no power and it should be possible to run that from a pseudo ground giving you +and - 12volts which will run just about any opamp just fine provided you use input and output coupling capacitors.

There are multiple complications with your current plan ... for example plus and minus 15 vac will end up being plus and minus 21 volts by the time you rectify and filter it...

Not to say your plan is no good... but to suggest an easier way.

 

[freemansteve]

Do you need any other voltages greater than 9V?
There are plenty of low-cost DC-DC up-converters for sale.

 

[dennnic]

I think for the 1.2A raspberry pi supply you might be better off looking for buck regulators -these are switching devices rather than linear and dissipate much less heat for the same power output. (Like class D amp instead of Class A)O therwise you've got at least the 7W or so I mentioned previously in the regulator. Pre or post capacitance won't help since the current still has to go through the regulator. On the other hand you can get a reasonably sized heatsinks. Here is one with 5.6C/W - which should limit temperature rise to around 42c above ambient - still quite toasty though.

After reading more and more about it, I decided to do just that. Use a stock smps for raspberry and save the transformer for a linear PSU for a spdif converter board on top of the pie.
It shouldn't draw more that 200mA.

By the way, if a regular needs to dissipate 1.5 - 2 watts, how big of a heatsink would be needed?

Can I make an alternative suggestion here...
I would use an SMPS brick power supply for this. Start with a 24 volt 2amp supply ... add regulators and additional filtering as needed.

You mentioned needing opamp supplies for a cluster of 7 opamps ... you likely only need about 100ma for that entire circuit. Opamps draw next to no power and it should be possible to run that from a pseudo ground giving you +and - 12volts which will run just about any opamp just fine provided you use input and output coupling capacitors.

There are multiple complications with your current plan ... for example plus and minus 15 vac will end up being plus and minus 21 volts by the time you rectify and filter it...

Not to say your plan is no good... but to suggest an easier way.

Thanks, but that might leave me with a problem of switching noise, that I've been trying to avoid from the beginning.

The discharge and charge rate of those caps that you suggest will be affected by the time constant(s) of the RC circuit(s) that is made. I'm not aware if using oversize caps is a design feature or not that is used for reducing the peak load on the linear regulator and even on the transformer as you mentioned previously. That is a question best answered by an electrical or electronics engineer.

I think I haven't been clear enough. I intended to use that large capacitance before the regular, 2-3 000 uF as a buffer capacitor and the same value cap for RC filter.

The reason for bringing that back:
If one 2200uF cap was followed by two RC filters, both of which used 2200uF caps, that would add up to 6600uF capacitance in total before the regulator.
If 1n400x diodes were to be used as a full wave rectifier, would there be too strong of a current for diodes (during startup or regular usage)?

 

[antcollinet]

By the way, if a regular needs to dissipate 1.5 - 2 watts, how big of a heatsink would be needed?

Look at the degree c per watt rating of the heat-sink (Thermal impedance). That will indicate how much temperature rise you will get at the surfaced of the device for a given power dissipation.

So for 2W, if your heat-sink is 20C per watt, you can expect a 40C rise (note this is affected by air flow and hence by the orientation of the heat-sink. The rating will assume vertical alignment of fins.

So if your room temperature is 20C then your device surface temperature is 60C. Then you need to worry about junction to case thermal impedance. For example a TO220 LM317 might have a junction to case resistance of 50C/W - so now your junction temperature is going to be 60C + 2x50C = 160C, exceeding the absolute maximum junction temperature of 150C.

If your power dissipation is only 1.5W, then you'd be at 1.5* (20 + 50) + 20 = 125C junction temperature.


You also have to consider your maximum ambient which may be much higher than 20C. If the power supply is in a box, how hot is it in the box. Also consider the thermal interface between the case and heat-sink. Normally a thin silicone pad or thermal grease is used.

 

[dennnic]

Look at the degree c per watt rating of the heat-sink (Thermal impedance). That will indicate how much temperature rise you will get at the surfaced of the device for a given power dissipation.

So for 2W, if your heat-sink is 20C per watt, you can expect a 40C rise (note this is affected by air flow and hence by the orientation of the heat-sink. The rating will assume vertical alignment of fins.

So if your room temperature is 20C then your device surface temperature is 60C. Then you need to worry about junction to case thermal impedance. For example a TO220 LM317 might have a junction to case resistance of 50C/W - so now your junction temperature is going to be 60C + 2x50C = 160C, exceeding the absolute maximum junction temperature of 150C.

If your power dissipation is only 1.5W, then you'd be at 1.5* (20 + 50) + 20 = 125C junction temperature.


You also have to consider your maximum ambient which may be much higher than 20C. If the power supply is in a box, how hot is it in the box. Also consider the thermal interface between the case and heat-sink. Normally a thin silicone pad or thermal grease is used.

Those are some huge numbers! In order to dissipate 7 watts, it would have required liquid nitrogen cooling solution.

One more thing regarding to junction to case parameter, since it poses the biggest factor.
Texas instruments have two variants of lm317 - KCT with junction to case of 50 and KCS with just 15. Analog instruments also offers two variants, both of which state junction to case of just 4. All of them packaged as To220.

Are those significant differences for real, or is it just a marketing gimmick?

 

[Ingenieur]

My Bad, +/-15v secondaries are rated at 1 amp, the load on them would be just 7 op amps.
There's both a thermal fuse inside (forgot it's rating) and a thermal switch glued on the side of the transformer.

Reason for all of this writing really is - I've read on several sources that secondary should supply at least twice the required current, or, preferably, three times more.

If 9v secondary would allow for 3.8 amps, then 1.2 amps load is slightly above the 1/3rd rule.
I'm wondering if to add a large capacitor storage before and after the regulator (average usage should be around 700-900 mA), or to connect both 9v secondaries together and connect two regulators to that. (Or maybe three. That 100 mA load, that's the only one that actually needs linear power supply)

What is the ' 1/3 rule ' ?

All loads are AC?
Why are regulators required? And caps?

It is a 4 winding xfmr?
1 primary
3 secondary , 1 x 30 center-tap, 2 at 9 VAC
1 x 30 + 2 x 3.8 x 9 = 100 VA

Strange configuration
Are the 2 x 9 opposite polarity? 180 out of phase?

A xfmr will only have Vdrop based on load.
At full load <3%, 8.7 VAC on the 9 tap
It is linear, 9 at no load, 8.7 at FLA

I assume it will be a bit high at no load and likely ~9 at full.

Just use as is imo

 

[dennnic]

What is the ' 1/3 rule ' ?

All loads are AC?
Why are regulators required? And caps?

It is a 4 winding xfmr?
1 primary
3 secondary , 1 x 30 center-tap, 2 at 9 VAC
1 x 30 + 2 x 3.8 x 9 = 100 VA

Strange configuration
Are the 2 x 9 opposite polarity? 180 out of phase?

A xfmr will only have Vdrop based on load.
At full load <3%, 8.7 VAC on the 9 tap
It is linear, 9 at no load, 8.7 at FLA

I assume it will be a bit high at no load and likely ~9 at full.

Just use as is imo

I've ordered it as a 9v center tapped (+9/0/-9), I've just reorganized wires to get one 9volt secondary. From there, the idea was to use multiple devices with separate regulation.

Edit: 1/3rd - I was told to chose the transformer, which secondary supplies three times the maximum current that would be used.

 

[egellings]

Separate regulators will also provide isolation between the devices powered by the transformer. Interactions between devices, if it happens, could degrade their performance.

 

[Deleted member 46664]

Thanks, but that might leave me with a problem of switching noise, that I've been trying to avoid from the beginning.

Unless you are buying the 50 cent specials from Ali you really won't have to worry about that much. For relatively low currents 100uf in parallel with a .1uf behind each regulator is usually adequate.

 

[antcollinet]

Those are some huge numbers! In order to dissipate 7 watts, it would have required liquid nitrogen cooling solution.

One more thing regarding to junction to case parameter, since it poses the biggest factor.
Texas instruments have two variants of lm317 - KCT with junction to case of 50 and KCS with just 15. Analog instruments also offers two variants, both of which state junction to case of just 4. All of them packaged as To220.

Are those significant differences for real, or is it just a marketing gimmick?

They'll be real. Important parameters for a design engineer. It'll be to do with the internal construction, and how well the device is coupled thermally to the case. I'm guessing the higher thermal resistance parts (which I chose to make a point) are older.