Depends on what exactly you are planning to measure.
If you are talking about power supply hum = residual AC component on the DC rails, the approach is quite simple. Just model the whole power supply including the power transformer and use a sine wave as input that equals your local mains. Keep in mind that LTSpice defines a sine wave via amplitude and frequency, so for 240VAC 50Hz mains you need to use a voltage source set to sine, 50Hz, 240*1.41=338V.
Compare the traces at the transformer secondary, behind the rectifier, after the first filter element and so on to learn about the power supply ripple.
Rundmaus
If you are talking about power supply hum = residual AC component on the DC rails, the approach is quite simple. Just model the whole power supply including the power transformer and use a sine wave as input that equals your local mains. Keep in mind that LTSpice defines a sine wave via amplitude and frequency, so for 240VAC 50Hz mains you need to use a voltage source set to sine, 50Hz, 240*1.41=338V.
Compare the traces at the transformer secondary, behind the rectifier, after the first filter element and so on to learn about the power supply ripple.
Rundmaus
Thanks for your reply, as i have read through several thread, people often measure power supply noise, ripple, rejection and output impedance. So to judge if a power supply is good or not, what's the best parameter? noise using .noise command or ripple rejection using ac analysis?
If a power supply is good or not depends on the requirements of the circuit it is powering - a good power supply for a PP amp might be inadequate for a SE circuit due to poorer ripple rejection and so on.
Rundmaus
Rundmaus
I suspect simulation is a "poor" tool for measuring noise generated by a PSU because there are many many factors that aren't figured in the sim. For example a cap placed across the rail will kill all the noise because the cap in simulation is "perfect", the same for the connecting leads etc.
Where simulation might be of use in testing how a discrete design responds for example to a pulsed or dynamic load. That's easy to set up using a FET and resistor as load and feeding the gate with a pulse waveform.
The reality of any PSU is that what is measured at its output and what is measured a few cm away down a wire or piece of print can be very very different.
Where simulation might be of use in testing how a discrete design responds for example to a pulsed or dynamic load. That's easy to set up using a FET and resistor as load and feeding the gate with a pulse waveform.
The reality of any PSU is that what is measured at its output and what is measured a few cm away down a wire or piece of print can be very very different.
I'd agree with Mooly that it is difficult to represent 'real world' noise in PSU simulations. But I think it can be very useful for the evalutation of residual ripple and filter chain design.
For me, it was also useful to model the response of the PSU to low-frequency mains variations, which can be done easily by appropriately modifying the voltage source used for the mains.
Rundmaus
For me, it was also useful to model the response of the PSU to low-frequency mains variations, which can be done easily by appropriately modifying the voltage source used for the mains.
Rundmaus
Reality check: So far, we don't even know what sort of power supply he's talking about - unregulated / linear regulater / SMPS.
I'm talking about discrete power supply like this one:
Now, i think i already know how to do it 😀. My next question is the load is going to be about 500ma, regulate 36V to 24V, that mean pass transistor will have to dissipate about 6W so can a single BD139 with big heatsink do the job or should i pararrel 2?
Now, i think i already know how to do it 😀. My next question is the load is going to be about 500ma, regulate 36V to 24V, that mean pass transistor will have to dissipate about 6W so can a single BD139 with big heatsink do the job or should i pararrel 2?
Yes, i haven't seen any circuit that more than 200ma pass through the bd139, i can't find the d44h11 at my country and using other transistor like tip41 make the ripple rejection much worse so i will stick with parrareling 2 bd139.
There's a slight problem with the schematic in post 9. The constant current source based on Q1 looks like it's only going to give about 4mA. If you want 500mA output, you're relying on Q2 to have a current gain of over 100, which is pushing your luck. OTOH, if you increase Q1's current too much, you may exceed the opamp's current sinking ability.
I have read that for ib should be about 1/20 of ic, the load in fact is less than 400ma, so is the 20ma ccs too much? The simulate work well with 40ma ccs.
I am not sure that feeding the opamp from the output is a good idea in this case: at first sight, it seems to eliminate the PSRR issues, but the auxiliary biasing that it requires could well bring more noise than using directly the raw supply (a good opamp has an excellent PSSR), and it leads to difficult tradeoffs.
You should try other options
You should try other options
- Status
- Not open for further replies.