# Dual power supply with two different amperages

I'm new at electronics so please bear with me. I'm trying to build a power supply for an old device that needs +5v at 3a and -5v at 220mA. Here's my circuit:

What I have is: a center-tap transformer that produces two 12v currents at up to 4a, which feeds into a bridge rectifier (Vishay PB5010), which outputs it as about ±14V DC. After 1000µF smoothing/decoupling capacitors, the current goes to a pair of voltage regulators, LM1084 (+5v 5a) and LM7905 (-5v 1.5a), then two 10µF decoupling capacitors, an indicator LED, and then to the device.
When I test it with a multimeter, it tells me that the voltage is exactly right from both outputs. However, when I plug the device in, I get the -5v at 220mA, which is correct, but then only 500mA from the +5v. I'm able to plug a +5v-only supply to the device and it will draw all 3a that it needs, so the problem must be in my circuit.
My thoughts ahead of time are:
-I probably don't understand the datasheet on the bridge rectifier; the one I got is rated at 4a regulated current, up to 1v at 22.5a, which I assumed meant the maximum amperage would go down proportionally as the voltage increased, to just about 4a at 5v, but it may just not mean what I thought it would.
-My understanding was that using a full-wave rectifier would allow the circuit to use the full 4a that the transformer can output, but that may not be true.

I've looked all over for answers to these questions but they're so specific that I haven't had any luck.
In brief my question is: How do I get 3a from the +5v?

• looks like a bottleneck from the transformer. What is the electrical specification of the transformer that you are using? Aug 19, 2020 at 0:12
• 1000uf on the +5V side is a bit small. Try adding more in parallel. Aug 19, 2020 at 0:32
• There are tons of linear designs here. Your 24V transformer shud be only 12V so you get +/-7V the losses are frying the regulator. Aug 19, 2020 at 1:17
• Or +/-8 V which is at rated load otherwise +10% no load. 5A is a 1 Ohm load 25W while regular dumps 35W from 14 to 5V. ,!! , RC =Tau 1*1000mF=1ms won’t work well with 8ms as it decays below the 7V regulator input and then you get pulsed low voltage and current out. Aug 19, 2020 at 1:30
• If this were me, I'd focus on the 5 V, 3A supply rail and then use an inverter to get the -5 V, 220 mA rail. And, that way you make good use of that transformer's secondary (less mass, lower cost.)
– jonk
Aug 19, 2020 at 2:48

If you want 3amps from the +5V regulator, the input voltage to that regulator should at every moment remain above +5v, plus some more because the regulator needs some overhead. The data sheet for LM1024 says that 1.5V of overhead is required when 5A flows - let's use that figure.
So input voltage to the regulator must always be above 5+1.5=6.5V

When 3A flows, the diodes charge up the big capacitor in a rather short time period near the peak of the 60 Hz wave...actually it is a 120Hz waveform because you're using a 4-diode full-wave rectifier. So assume the capacitor charges all-at-once, then discharges @ 3A for a period of 1/120 seconds....0.0083s
You've measured the unloaded DC output voltage near 14V. This seems like plenty for a regulator chip that only requires 6.5V at its input. But 3A pulled out of the capacitor depletes its voltage rather quickly...too quickly. You need a big reservoir of charge, and that means a big capacitor.

You can work out minimum capacitance required from $$\I=C \frac{dV}{dt}\$$ where I=3A, dV=(14-6.5) and dt=0.0083 seconds. Find C in units of farads. Your C of 1000uF is too small.

• Oh thank you! I was wondering if it had to do with the capacitance, that was something I was having trouble finding how to calculate. If I'm doing the math right (which I may well not be) it sounds like I'll need about 3320µF minimum. I'll try that when I get a chance! Aug 22, 2020 at 17:30
• You should be OK with 1000uf on the -5V side (because max current is lower). On a different note - do a thermal analysis - data sheets show max current 5A assuming that you have perfect heat-sinking. These chips can easily overheat if your heatsink is inadequate. Aug 22, 2020 at 19:18

The transformer secondary voltage (12V - 0 - 12V) is too high and would require the regulator to dissipate close to 50W for only 15W output. Should the heat-sinking not be adequate, the regulator would limit the current to lower the dissipation.

The voltage is to be decided after factoring in the 'line regulation' (mains voltage variation), the 'load regulation' (transformer winding and diode drops on load) and the drop-out voltage of the regulator.

With the LM1084 having a drop-out voltage of 1.5V, a 7V - 0 - 7V transformer secondary would be good enough.

The filter capacitor value is too low at 1000 μF. 4500 μF would be required for a 3A supply.