Bi-regulated power supply with no end capacitor

[HendriXML]

Hi,
I'm trying to design a very stable power supply without using end caps. The reason for this is the possibility to add a short-circuit protection, that activates without discharging an end cap.
It also needs to be stable driving inductive loads. The attached circuit seems to be working, but still has some 40 mV noise on the output. Propably because of the low differential voltages (mV) it is regulating on.
So I would like some advise on how to keep the noise down.

The circuit needs some delicate tuning on some trimmers to get it working correctly, so I already know it is not for mass manufacturing . But as a beginner, I find this project very useful to explore some practical limits and get a better understanding of things.
First I shall explain my thoughts when I came up with the design. There're 2 sections: one for constant voltage and one for constant load current regulation.

Section 1: constant voltage
First of all I wanted to separate power supply variation and load variations (including inductive load voltages). Sensing resistors R3 and R6 are used for that.

• R3 is sensing the Zener diode (D1) current and the load current.
• R6 is sensing a constant current (Q3) and the load current.

The op amp U1 is used to regulate a balance between the voltage drops.

Voltage variations on the input are sensed by R3 and regulated down by Q1.
Voltage variations/spikes on the output are sensed by R6 and regulated by Q2 in section 2.

Section 2: constant load current
In this section the stable voltage of 5.1V is regulated down to 5.0V by controlling the current through R6 to about 450mA.
Transistor Q4 ensures that Q1 is initially conducting enough to ensure that a balance between the constant current and the Zener current can be regulated.

--I'm testing on a bench power supply, so no ripple capacitors on board yet.

Feel free to comment.

Thanks in advance

[HendriXML]

I powered the circuit to a adapter with 2 x 2200 uF ripple capacitors and did some further experimenting. Bringing the zener current down made the output noise better.

I now got a ripple of 11 mV peak tot peak. That is with a nasty inductive load, that gave my lineair bench power supply a lot more noise (80mV peak tot peak).

For me that is a winner winner chicken dinner!

After seeing this EEVBlog YouTube stuff about ripple and noise measurement.
https://youtu.be/Edel3eduRj4
I would like withdraw the measurements I’ve done  . Maybe I’ll do some proper measurements following the advise of Dave!

[HendriXML]

If I stall the motor the noise becomes even less. Like 5 mV peak to peak.

[Rerouter]

Put an excentric mass on your motor shaft if you want a truly horrible test load. Inductance is a far more horrible problem when the current is constantly changing.

[HendriXML]

Good idea! I have to figure something out then. But the circuit should handle that, because the currents remain quite constant. The second section in the schematic ensures that.

That part alone did wonders to my bench power supply output but that one has a end capacitor which makes dampening a lot easier.