2
\$\begingroup\$

I'm working on a constant current power supply and as a first step designed the following rectifier with a filter. I know the bottom rectifier is inactive, but with couple of switches the circuit should be able to act as two separate power supplies, though with orders of magnitude higher noise of ~40mV.

enter image description here

Time domain transient simulation in PSpice for TI with general settings, 3s run time and 3us step size predicts a ripple of ~200uV:

enter image description here

As far as my understanding goes, the effectiveness of a capacitance multiplier depends on the gain of the transistor (maybe even throw in a Darlington,) which I did not specify anywhere and don't even know how to, so I'm having doubts about the accuracy of the simulation. Also, I'm having trouble determining what specifications the transistors should have, like maximum collector current, power dissipation, etc.

\$\endgroup\$
7
  • 2
    \$\begingroup\$ please draw your schematics +ve to the top, it makes it so much easier to communicate to others \$\endgroup\$ Commented Nov 8, 2023 at 11:01
  • 1
    \$\begingroup\$ If you didn't specify the transistor, it probably gave a default beta of 100, which is reasonable for small signal transistors. Hint, if you're designing a constant current supply, there's no need to bust a gut getting low voltage ripple from your rectifier. Those capacitance multipliers are unnecessary and insufficient. If your current source stages are to handle load transients, they could easily be made to have good enough PSRR to handle a lot of supply ripple. \$\endgroup\$ Commented Nov 8, 2023 at 11:04
  • \$\begingroup\$ @Neil_UK I looked at the properties of the transistor, but did not find anything akin to beta or gain or even any number except for 0W BiasValue Power. Also, what do you mean the capacitance multipliers are insufficient? \$\endgroup\$ Commented Nov 8, 2023 at 11:10
  • \$\begingroup\$ how much peak to peak input ripple will the cap multipliers handle before their output becomes compromised? \$\endgroup\$ Commented Nov 8, 2023 at 12:22
  • \$\begingroup\$ @Neil_UK I don't know? \$\endgroup\$ Commented Nov 8, 2023 at 12:30

3 Answers 3

Reset to default
1
\$\begingroup\$

Also, I'm having trouble determining what specs should the transistors have, like maximum collector current, power dissipation, etc.

A very common industry rule of thumb for long-term reliability, and a requirement in some companies, is 2:1. Over-rate components by 2:1.

If the continuous power dissipation in a resistor, diode, or transistor is 1 W, use a 2 W part minimum.

If the peak collector current in a transistor is 1 A, use a 2 A part minimum.

Same for capacitor voltage, collector-emitter / drain-source voltage, diode reverse voltage, diode current, etc.

Sometimes this can't be done for supply / cost / space reasons, and of course there are exceptions. For example, X- and Y-rated power line filter capacitors are rated for 275 V, but are designed to operate continuously at 240 V.

\$\endgroup\$
0
\$\begingroup\$

Your concerns are reasonable. Realistic values can be input by selecting a library model file, or downloading and importing one. If you need help, check the manual or get support for your simulator.

Mind there are limits on the accuracy of such models, and whether the manufacturer fitted the real data accurately; it can pay to set up tests and check them yourself. This is particularly true when doing more unusual things, especially high frequency, inverted operation, exact switching behavior*, etc. This application should only need a few realistic parameters though; RC, RE, BF, VA, IS's and N's. More if you intend to test behavior over temperature, and more accurate noise levels (i.e. where/if thermal noise applies).

*Not so much a fault of the modeling or fitting process, but SPICE itself uses a fairly crude approximation for reverse recovery (and by extension, BJT charge storage). This is relevant for short-duration pulses, comparable to or shorter than the recovery time of the junction in question; so it causes some error in switching events, but for general purposes isn't too awful.

\$\endgroup\$
0
\$\begingroup\$

What depends on beta, is mostly the output impedance. As you have a constant load, this is not a concern.

What is a concern however, is that your circuit provides no low frequency PSRR at all. So while the mid-high frequency ripple is suppressed well, even the tiniest drift in the line voltage will appear as change in current through your load.

Is your mains voltage really stable to within less than 1 mV? No..

\$\endgroup\$
2
  • \$\begingroup\$ I don't have a constant load, it is going to slowly increase once the voltage is applied. How do I simulate those drifts and what frequency and amplitude of them should I expect from mains? \$\endgroup\$ Commented Nov 8, 2023 at 15:19
  • \$\begingroup\$ @beardeadclown check the internet for mains power tolerance in your region. I think 5% is not unheard of. And that would have a commensurate effect, i.e. 5 %, on your output current. At 320 V, this would be a 16 V change, dwarfing the rejected 50 Hz ripple. \$\endgroup\$ Commented Nov 8, 2023 at 15:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.