How to simulate PSUs in PSUD2?

Help Support GroupDIY:

rock soderstrom

⚽️⚽️⚽️⚽️⚽️
Joined
Oct 14, 2009
Messages
1,171
Location
Berlin
Hello, I knew the program PSUD2 for quite some time and I have also used it a few times in the past. Mostly I just build my power supplies based on experience and trial and error.

It would be nice to do the dimensioning in advance through simulation.

How do you guys do this? Do you use PSUD2? How do you do it right?

As an example, I have a transformer with a secondary voltage of 240V (30mA) and want to use it to power a single channel microphone amplifier. Let's say the circuit (Gates SA70) draws 10mA and I want to filter the B+ voltage with a CRCRCRC filtering without regulation. I want the DC voltage after filtering to be 240V DC.

How do I set the load resistor (R4)? My gut tells me R4 should define the expected current flow, so my assumed 10mA, right?

Is this correct or nonsense?
 

Attachments

  • 2021-03-09.png
    2021-03-09.png
    173.5 KB · Views: 17

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,771
Location
Norfolk - UK
Your load is 10mA at 240V which is just a resistor of 240 / 10 K or 24K.

Your transformer has 240VAC secondary so the peak voltage on the first capacitor will be 240 x 1.414 = 340V

You need to drop this to 240 volts so the total volts drop is 340 -240 = 100V

The total resistance needed is therefore 100 / 10 K = 10K. You need three equal resistors so each one is 3K3.

You do not really need a simulator for such a simple supply but it is useful because it shows you the ripple voltages and inrush currents.  (Although you can calculate these quite easily also)

Cheers

Ian
 

emrr

Well-known member
Joined
Apr 12, 2006
Messages
7,875
Location
NC, USA
Some spec updates:  SA-70's draw roughly 5mA at 275VDC.  I haven't run them higher, but you certainly could, given the voltage drops in the dividers, 100V on each plate at 240V B+.  B+ isn't at all critical, and max output is already low by modern standards, so I wouldn't be worried about 300V B+, or higher.  If I had a clean 400V B+, I wouldn't worry about using that so long as cap and resistor ratings were sufficient.  FWIW. 
 

Tubetec

Well-known member
Joined
Nov 18, 2015
Messages
3,124
I just recently visited Duncans amps to download PSUDII .
Seems like now its no longer free , luckily I had an archived version stored away .

I sometimes use it to work backwards , so basically I have a transformer I dont know the spec of , add a rectifier a cap and a load of a known value , now you can re-jig the transformer values until it lines up perfectly with what your seeing .
 

rock soderstrom

⚽️⚽️⚽️⚽️⚽️
Joined
Oct 14, 2009
Messages
1,171
Location
Berlin
ruffrecords said:
Your load is 10mA at 240V which is just a resistor of 240 / 10 K or 24K.

Your transformer has 240VAC secondary so the peak voltage on the first capacitor will be 240 x 1.414 = 340V

You need to drop this to 240 volts so the total volts drop is 340 -240 = 100V

The total resistance needed is therefore 100 / 10 K = 10K. You need three equal resistors so each one is 3K3.

You do not really need a simulator for such a simple supply but it is useful because it shows you the ripple voltages and inrush currents.  (Although you can calculate these quite easily also)

Cheers

Ian

Thanks Ian for the quick lesson in calculating a PSU! Basically I was aware of this, but still very informative.

My focus was more on simulation, out of pure interest of the possibilities. If I take your correct numbers I come to completely wrong results with the simulation :eek:, surely the problem sits in front of the display. I will try it again today....

It would be interesting to have a short lesson from you regarding the calculation of the resulting ripple and what values you are aiming for in such a design. (MicPreAmp). Of course, only if your schedule allows it!
 

rock soderstrom

⚽️⚽️⚽️⚽️⚽️
Joined
Oct 14, 2009
Messages
1,171
Location
Berlin
Tubetec said:
I just recently visited Duncans amps to download PSUDII .
Seems like now its no longer free , luckily I had an archived version stored away .

Strange, I was still able to download it (V2.20 Build 74) for free, yesterday. http://www.duncanamps.com/psud2/download.html
 

abbey road d enfer

Well-known member
Moderator
Joined
Jan 22, 2008
Messages
13,030
Location
Marcelland
rock soderstrom said:
My focus was more on simulation, out of pure interest of the possibilities. If I take your correct numbers I come to completely wrong results with the simulation
Calculating PSU's in broad strokes is easy. You must remember dV=I/C.dt. That's the peak ripple of a single cap filter. The more caps you add the more complicated is the calculation, if you want precise numbers. But, if you calculation says you need 56uF, what are you gonna do? I would put a 100uF, because electrolytic caps vary with age and there is not much choice, since most manufacturers are using the E3 table of values 1, 2.2 and 4.7.
Now if you want to dig deeper, use a simulator. I use LTspice, that is capable of much more than calculating PSU's.
I just downloaded the new version of PSpice for TI, which seems to have a steeper learning curve, but has some interesting features (mainly of interest for purchasing and modelling).
 

rock soderstrom

⚽️⚽️⚽️⚽️⚽️
Joined
Oct 14, 2009
Messages
1,171
Location
Berlin
EmRR said:
Some spec updates:  SA-70's draw roughly 5mA at 275VDC.  I haven't run them higher, but you certainly could, given the voltage drops in the dividers, 100V on each plate at 240V B+.  B+ isn't at all critical, and max output is already low by modern standards, so I wouldn't be worried about 300V B+, or higher.  If I had a clean 400V B+, I wouldn't worry about using that so long as cap and resistor ratings were sufficient.  FWIW.

Thanks Doug for the real specs and your suggestions! I had only chosen the 10ma because it is easy to calculate with. Today I will start to populate my real SA70 testrig after some experimenting with a rats nest build. I like this amp and the sound of that octal tubes! I have named my project ST70, it is a 1HU, one channel version with some extras. Thanks again for your help and inspiration.
 

Attachments

  • 2021-03-10.png
    2021-03-10.png
    297.4 KB · Views: 11

rock soderstrom

⚽️⚽️⚽️⚽️⚽️
Joined
Oct 14, 2009
Messages
1,171
Location
Berlin
abbey road d enfer said:
Calculating PSU's in broad strokes is easy. You must remember dV=I/C.dt. That's the peak ripple of a single cap filter. The more caps you add the more complicated is the calculation, if you want precise numbers. But, if you calculation says you need 56uF, what are you gonna do? I would put a 100uF, because electrolytic caps vary with age and there is not much choice, since most manufacturers are using the E3 table of values 1, 2.2 and 4.7.
Now if you want to dig deeper, use a simulator. I use LTspice, that is capable of much more than calculating PSU's.
I just downloaded the new version of PSpice for TI, which seems to have a steeper learning curve, but has some interesting features (mainly of interest for purchasing and modelling).

Thank you! Very interesting. Maybe I should take a look at LTSpice after all.

 

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
13,771
Location
Norfolk - UK
rock soderstrom said:
It would be interesting to have a short lesson from you regarding the calculation of the resulting ripple and what values you are aiming for in such a design. (MicPreAmp). Of course, only if your schedule allows it!

As Abbey says, you can work out the base level ripple from C.dV = i.dt

Where C is the value of the cap right after the rectifier, dV is the ripple voltage (peak to peak), i is the load current and dt is the period of the waveform. Abbey rearranged it into a more useful form:

dV = i x dt / C which can itself be rearranged into the even more useful form of:

dV = i / (f x C)

where f is the ripple frequency (usually twice the mains frequency for full wave rectifiers)

In your example, with a 10mA load and a 47uF capacitor, the ripple will be 2.13V pp.

The following RC stages each reduce the ripple by an amount equal to:

2 x PI x f x R x C

which for a 47uF cap and a 3K3 resistor gives a reduction of 97 times or very nearly 40dB. So three stages like this will reduce the ripple by about 120dB. I always rememeber 60dB converts volts to millivolts and another 60dB converts millivolts to microvolts. So the ripple should be less than a couple of microvolts.

Cheers

Ian

 

Tubetec

Well-known member
Joined
Nov 18, 2015
Messages
3,124
Not sure why the site now diverts to Group.io then back to the original homepage , anyway yeah PSUD is still available for free.
 

rock soderstrom

⚽️⚽️⚽️⚽️⚽️
Joined
Oct 14, 2009
Messages
1,171
Location
Berlin
ruffrecords said:
As Abbey says, you can work out the base level ripple from C.dV = i.dt

Where C is the value of the cap right after the rectifier, dV is the ripple voltage (peak to peak), i is the load current and dt is the period of the waveform. Abbey rearranged it into a more useful form:

dV = i x dt / C which can itself be rearranged into the even more useful form of:

dV = i / (f x C)

where f is the ripple frequency (usually twice the mains frequency for full wave rectifiers)

In your example, with a 10mA load and a 47uF capacitor, the ripple will be 2.13V pp.

The following RC stages each reduce the ripple by an amount equal to:

2 x PI x f x R x C

which for a 47uF cap and a 3K3 resistor gives a reduction of 97 rimes or very nearly 40dB. So three stages like this will reduce the ripple by about 120dB. I always rememeber 60dB converts volts to millivolts and another 60dB converts millivolts to microvolts. So the ripple should be less than a couple of microvolts.

Cheers

Ian

Perfect!!! Thank you so much! I will apply my new knowledge today. I will compare the math with the simulation and the real circuit after I finish my test setup. I am very curious...
 

trobbins

Well-known member
Joined
Jul 1, 2011
Messages
278
Location
Australia
PSUD2 is certainly the easiest way to 'calculate' load DC voltage and ripple voltage at the load.  In your original PSUD2 plot those load levels can be shown by using a 'reporting delay' of say at least 1 sec to see estimated waveforms and plot levels, and the 'Result' section on the LHS can be expanded to see levels like pk-pk ripple levels.  If the voltage across, or the current through the load R4 are not what you want then change the value of R4 (but note that R4 = the required load voltage divided by the required current being drawn).  Obviously if you need a higher load voltage for a given R4 then reduce one/all R1-R3.

You need to insert the measured values for transformer voltage and winding resistance to get an valid estimate - I find it is best to use the 'Source Impedance calculator".
 

rock soderstrom

⚽️⚽️⚽️⚽️⚽️
Joined
Oct 14, 2009
Messages
1,171
Location
Berlin
trobbins said:
PSUD2 is certainly the easiest way to 'calculate' load DC voltage and ripple voltage at the load.  In your original PSUD2 plot those load levels can be shown by using a 'reporting delay' of say at least 1 sec to see estimated waveforms and plot levels, and the 'Result' section on the LHS can be expanded to see levels like pk-pk ripple levels.  If the voltage across, or the current through the load R4 are not what you want then change the value of R4 (but note that R4 = the required load voltage divided by the required current being drawn).  Obviously if you need a higher load voltage for a given R4 then reduce one/all R1-R3.

You need to insert the measured values for transformer voltage and winding resistance to get an valid estimate - I find it is best to use the 'Source Impedance calculator".

Thanks a lot for your input!👍

I am still on topic and I hope I will find a little spare time today to go ahead with my comparission betwen simulation and the real build.
 
Top