Maker Pro
Maker Pro

Need help with a PSU schematic

duke37

Jan 9, 2011
5,364
Joined
Jan 9, 2011
Messages
5,364
Looks OK to me but you may like to put a double gap on the bottom track.
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
Yea...I will need to spread things out a bit as the rectifiers are wider than what DIYLC will do. They actually cover three holes wide. I also plan to, once I get the regulator sections done (have to allow for the heat sinks), open it up in Photoshop and flip it over (left to right) and resize to the actual Veroboard dimensions so I can mark what strips need to be broken and where.
 
Last edited:

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
Here's a simulation of the exact circuit you've drawn up, excluding the regulators. I'm describing the high-voltage rail as "bootstrapped" because BR1 uses the transformer secondary voltage directly, and BR2 uses the transformer's output voltage again, to add more voltage on top of that, which in some sense can be described as bootstrapping, in the electronic sense of the word.

attachment.php


I had to draw the bridges out as discrete diodes. LTSpice doesn't have a bridge rectifier as standard; I found one in a library but I couldn't figure out how to include it easily, so I gave up.

I've added RL1, RL2 and RL3 as load resistors to draw 0.5A from each low-voltage rail, and 0.1A from the high-voltage rail.

It's actually pretty important to know the maximum current you will be drawing from each of these rails. It affects the capacitor values you will need. The value of C11 has implications for C9 and C10 as well, as I'll explain below.

Here's a graph of the voltages on V1, V2 and V3.

attachment.php


You can see that there's a fair amount of ripple on all the rails. This could be reduced by increasing the relevant output smoothing capacitor, but as long as the trough in the ripple is comfortably high enough (including tolerances on transformer ratio, AC mains supply voltage, and anything else), you don't have to.

attachment.php


The graph above shows the AC voltage at point "S1" (the top end of the secondary), and the CURRENTS through C9 and C10. This is important, because current only flows in these capacitors for a short time, as you can see.

The upward spikes on the blue and red traces represent C9 and C10 (respectively) being CHARGED from the transformer. The downward spikes represent them being DISCHARGED into C11.

As you can see from the scale on the right side, the positive and negative peaks are both about 0.5A. That is with a load of only 100 mA on the V3 rail. So with the component values you originally specified, the peak positive and negative ripple currents in C9 and C10 are about five times the load current on the V3 output.

This is significant because ripple current can (and does) damage electrolytic capacitors in the long term, and you should investigate the ripple current specificaitons of the capacitors you use for C9 and C10.

I started writing this post yesterday, but I'm not going to be able to finish it for a couple more days, so I thought I should post what I have so far. I will run some more simulations with different values of C9/C10 and C11 to see what would be the best approach to minimise stress on C9/C10. I will probably recommend adding some small-value resistors in series with them (something in the range 2~10 ohms or so) to reduce the peak currents.

In the meantime it would be helpful if you could specify the maximum expected current drain on the three voltage rails.
 

Attachments

  • schematic.gif
    schematic.gif
    10.4 KB · Views: 155
  • voltages.gif
    voltages.gif
    8.5 KB · Views: 174
  • currents.gif
    currents.gif
    19 KB · Views: 161

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
WOW! :D That is much more than I can do with LTspice. THANKS! for all the effort.

There is a blog I found when I first started this project that dealt specifically with this particular Soundcraft mixer's original PSU. http://www.neufeld.newton.ks.us/electronics/?p=498 Ultimately, the FET needed to repair the original PSU is no longer made and I couldn't find a work alike. This blog is also where I found a rough schematic for a linear PSU that I based the one here on.

From the blog, the current drain should be:

230mA from the +15V rail
260mA from the – 15V rail
20mA from the +48V rail

The blog entry also lists the resistors needed to load the PSU the same amount. Those are, respectively:

(68R at least 4 watt)
(56R at least 5 watt)
(2K4 at least 1.5 watt)

And then there is the thread over at DIYAudio where this schematic developed into much more. http://www.diyaudio.com/forums/power-supplies/239750-please-check-schematic.html

Much of what they were saying was going over my head so I came here in the hopes that I may understand better what is being said. I can follow a schematic (this goes here and that goes there) and can build but, when it comes to design, I tend to get lost.

Again to everyone, THANKS for the effort and input.
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
Just thought I would throw these up as well. These are two schematics for two other Soundcraft mixers that are larger than mine (more channels) but have similar voltage requirements. They both show pretty simple PSUs
 

Attachments

  • soundcraft_spirit_lx7.pdf_1.png
    soundcraft_spirit_lx7.pdf_1.png
    19.9 KB · Views: 110
  • CPS150.jpg
    CPS150.jpg
    95.3 KB · Views: 113

duke37

Jan 9, 2011
5,364
Joined
Jan 9, 2011
Messages
5,364
This is similar to my simulation but much neater.

Differences
1. I used 50Hz not 60Hz
2. I put 10R in the transformer secondary, not 0R2.
3. I loaded the low voltage with 100R, not 48R.
4. I loaded the high voltage with 2000R not 612R
5. I Used two 470uF caps on the low voltage output totaling about 1mF as recommended.
6. I did not simulate the negative output.

The results seem to be similar.
I do not see the need for D9 or D10.
How do you get such a nice circuit digram with reasonably sized text?
How do you capture the output graphs? I used screen capture and Irfanview which is a bit tedious.
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
There is a blog I found when I first started this project that dealt specifically with this particular Soundcraft mixer's original PSU. http://www.neufeld.newton.ks.us/electronics/?p=498 Ultimately, the FET needed to repair the original PSU is no longer made and I couldn't find a work alike.
Is that the problem with the original power supply?

The STP7NB60FP is nothing special. It's an N-channel MOSFET, rated at 600V, 4.1A, 1.2 ohms Rds, in a TO-220 full plastic package.

It's flagged as obsolete now, but there are many suitable replacements available from Digikey. Have a look at:
http://www.digikey.com/product-detail/en/AOTF7S60L/785-1521-5-ND/3152620
http://www.digikey.com/product-detail/en/R6006ANX/R6006ANX-ND/2791493

From the blog, the current drain should be:
230mA from the +15V rail
260mA from the – 15V rail
20mA from the +48V rail
That's good to know, but I'm suspicious of the 48V rail figure. The current drain will depend on how many phantom powered microphones you're using. Do all inputs on the mixer support phantom power? Do you know how much current a typical phantom powered microphone draws? I would rather use a comfortably higher figure, such as 60 mA. I will use 60 mA for my subsequent simulations.
6. I did not simulate the negative output.
Right, and there's no need to. I just did that so the schematic would look complete.
The results seem to be similar.
I do not see the need for D9 or D10.
They're to protect C9 and C10 from negative voltage if the bootstrapped rail is excessively loaded.

Edit: You're right, there's no need for them. C9 and C10 are protected from reverse voltage by the combination of the diodes in the two bridge rectifiers.

How do you get such a nice circuit digram with reasonably sized text?
Standard text is font size "1.5". Right-click on any text and you can change the size from the drop-down list. I think your drawing just needs the components to be spread out more.
How do you capture the output graphs? I used screen capture and Irfanview which is a bit tedious.
Yes, it's tedious. I use the PrtSc key to capture the screen, paste that into Paint Shop Pro (a generic image editor), crop to the desired area, reduce to 16 colours and export as GIF. A common mistake is using .JPG files for diagrams; JPEG is only suitable for continuous-tone images such as photographs (that's what the P stands for, after all).

Using JPEG on images with sharp edges produces horrible compression artifacts (really easy to see if you zoom in); to avoid them, you have to use a high quality factor which yields a large file size. GIF or PNG are much more appropriate.
 
Last edited:

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
Is that the problem with the original power supply?

The STP7NB60FP is nothing special. It's an N-channel MOSFET, rated at 600V, 4.1A, 1.2 ohms Rds, in a TO-220 full plastic package.

It's flagged as obsolete now, but there are many suitable replacements available from Digikey. Have a look at:
http://www.digikey.com/product-detail/en/AOTF7S60L/785-1521-5-ND/3152620
http://www.digikey.com/product-detail/en/R6006ANX/R6006ANX-ND/2791493

I never could find, or have anyone be able to, answer that question. If I had, I probably never would have ventured down the linear PSU path. My original intent was to replace everything that could have possibly gone out on the original SMPS. The FET was the only hangup. The up side is that the original PSU was prone to going out so, a linear PSU, I'm sure, will be more reliable.


That's good to know, but I'm suspicious of the 48V rail figure. The current drain will depend on how many phantom powered microphones you're using. Do all inputs on the mixer support phantom power? Do you know how much current a typical phantom powered microphone draws? I would rather use a comfortably higher figure, such as 60 mA. I will use 60 mA for my subsequent simulations.

I only own one, at the moment, mic that requires 48v Phantom power. A quick look at it's specs shows that it draws "<3.0mA". The maximum mic inputs on the mixer is 8. But, I use a M-Audio 2496 2 channel PCI card to record with. Which means that, at the most, I would be using only 2 mics that require phantom power. I just checked another mic that I may get that requires 48v phantom power and it draws "2 mA typical", according to it's spec sheet.
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
OK that's good. I'll revise my load figure for the bootstrapped rail down to 40 mA. Some of that is for the regulator.

I see from the comments on Keith's blog that several people have had problems with that power supply. But it's going to be easier and more convenient to fix it, rather than building a new supply. Also, is there enough space inside the mixer for a linear supply?
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
At this point, I am committed to a linear PSU. I already have $60.00+ invested in parts. A new original PSU cost $90.00. My goals where to get out a bit cheaper, learn something and hope to have a cleaner PSU than the original SMPS.

There is space enough inside. Though, I am thinking that it may be better to build a 1u rack mount enclosure. Mostly to keep the AC out of the same housing as the mixer.
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
This is significant because ripple current can (and does) damage electrolytic capacitors in the long term, and you should investigate the ripple current specificaitons of the capacitors you use for C9 and C10.

Here is a screen shot from the datasheet for the caps. The highlighted line is what I have intended for C9 & C10. If I am interpreting the lower graph correctly as well as the data sheet, they are not up to the task.
 

Attachments

  • Panasonic specs.jpg
    Panasonic specs.jpg
    153 KB · Views: 128

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
A new original PSU cost $90.00.
I thought you were going to fix the old one, not buy a new one.
We can help you fix the old one. Start by describing the symptoms. Are there any obviously damaged components? Have you checked the flyback clamp zener?

Also, you could tell us where you're located. I assume you're in America? There's a place for that on your profile. So I guess Digikey is an appropriate source? Do you have a preferred supplier? I can look on their web site and suggest a suitable MOSFET.

Re the ripple current specification. The RMS ripple current according to my simulation is about 300 mA so you should be alright. I'll post again tomorrow.
 
Last edited:

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
Actually, on the original PSU, I removed a couple of obviously fried caps and the resistors next to them. I also pulled the FET to test, although, I couldn't get any consistent readings. Problem is that a couple very thin traces lifted in the process. While I know I can work around the lifted traces and repair the SMPS, it still has a reputation of failure and if the linear PSU should fail at some point, I know I can fix it on my own.

I have already spent $$ and received most of the parts from Mouser to build a linear PSU. From here, it's just small adjustments to the circuit.

I have used Digikey before but, I find Mouser's online catalog easier to use.

I saved the image of the schematic that you used in LTspice and am going to try and copy it. I am hoping to try different caps and see how it effects the circuit, and may learn how to use LTspice a bit in the process.
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
Finally!! After quite a few hours, I was able to get LTspice to do what I wanted. A good deal of trial and error along with stubborn persistence paid off. :) The schematic was not problem but getting it to simulate was the real headache.

So, I got to playing around with different values of caps for C1, C2, C9, C10 & C11 trying to address the concerns posted by KrisBlueNZ (I appreciate your effort much more now). Looking to bring down the amperage spikes (without dropping the voltage) from C9 & C10 that were hammering C11 as well as looking to bring ripple down across the +/-15vdc rails though C1 & C2.

What I found to work the best was to change C1 & C2 (Coincidentally, C1 & C2 on the test schematic) from 470uF to 1000uF.

C9 & C10 (C3 & C4 on the test schematic) were dropped from 47uF to 10uF. To drop them any more would have started dropping the voltage. I also changed these to bi-polar caps as it's my understanding that they will tolerate the abuse better. Though, 4.7uF looked good for the spikes, the voltage through them was to low.

After play with the values of C11 (C5 on the test schematic) I found that remaining at 470uF was the best compromise.

Attached are the test schematic as well as voltage and amp spike graphs.
 

Attachments

  • TEST SCHEM.jpg
    TEST SCHEM.jpg
    126.1 KB · Views: 183
  • TEST VOLTAGE.jpg
    TEST VOLTAGE.jpg
    64.3 KB · Views: 94
  • CAP SPIKES.jpg
    CAP SPIKES.jpg
    83.9 KB · Views: 113
Last edited:

davenn

Moderator
Sep 5, 2009
14,254
Joined
Sep 5, 2009
Messages
14,254
What I found to work the best was to change C1 & C2 (Coincidentally, C1 & C2 on the test schematic) from 470uF to 1000uF.

I did suggest that to you way way back in post #10 ;)

DAve
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
I did suggest that to you way way back in post #10 ;)

DAve

I know. That's why I tried it. :) I fact, I had already purchased them at the local Radio Shack (Rat Shack) the end of last week,
 

KrisBlueNZ

Sadly passed away in 2015
Nov 28, 2011
8,393
Joined
Nov 28, 2011
Messages
8,393
AFAIK, bipolar electrolytics won't tolerate ripple currents better than standard electrolytics. Also, lower value electrolytics have lower ripple current limits, so you may be chasing diminishing returns.

I may have overestimated the importance of ripple current. Although the current waveform looks nasty, with large peaks, it's the RMS ripple current that's important and at 47 uF the RMS ripple current is only about 220 mA, which isn't THAT bad. So 47 uF might be a good choice.

But if you have a working LTSpice simulation, you're in a good position to evaluate your options. I suggest you grab a few data sheets for good quality Japanese electrolytics (NCC, UCC, Rubycon) from Digikey so you can compare ripple current specifications for different values.
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
Playing around a bit more with LTspice. Mostly with C1 & C2 on the +/-15vdc rails. What I found is that increasing the values (like from 470uF to 1000uF) will decrease the AC ripple voltage but, will also increase the current spikes and beat on the caps even more. What I am learning is that I need to find a happy medium. I can lower the AC ripple voltage at the expense of C1 & C2 lifespan.

Looking at the specs from another Soundcraft linear PSU, they show noise at -68dBu on the low voltage rail and -80dBu on the 48 volt rail. This should be very realistic to achieve after looking over the regulator's datasheets.

Back to more simulations.......
 

solo2racr

Aug 21, 2013
142
Joined
Aug 21, 2013
Messages
142
Also....Just looked up the data sheet for the 47uF 63v Panasonics I have intended for C9 & C10. The ripple spec for them is 222mA RMS @60hz. Based on the sim you ran, KrisBlueNZ, and found 220mA RMS, your right. The 47uF should work just fine.

It's C1 & C2 that are getting really beat on.
 

duke37

Jan 9, 2011
5,364
Joined
Jan 9, 2011
Messages
5,364
Its me again Kris !

I have simulated a half wave version with a 1M resistor across the 47uF capacitor. uA thus means voltage. I can get down to 200R load before the voltage goes negative, then it is limited to about 1.5V i.e. 2 diode voltage drops. It seems to me that the diodes across the 47uF capacitors are not necessary but would do no harm.

Duke
 

Attachments

  • LowResistance.png
    LowResistance.png
    26.7 KB · Views: 93
Top