Thoughts on PSU

M Gregg · 11th April 2011, 09:22 PM

Any thoughts?

I would be interested in your thoughts on these two circuits!
I know in rush is an issue on one of them. This is covered with soft start!

Regards
M. Gregg

mastodon · 12th April 2011, 08:12 AM

The first one looks like the more affordable option, although you'll have to make sure the diode bridge doesn't get fried due to the high capacitance following it. I'd personally prefer this one vs. the tube rectified, double-choke power supply because the latter is relatively expensive and occupies a lot more space.

DF96 · 12th April 2011, 09:47 AM

Both suffer from ringing in the LC part because they are insufficiently damped. Fortunately the final RC section attenuates this. What are they for?

M Gregg · 12th April 2011, 03:30 PM

Quote:

Originally Posted by DF96

Both suffer from ringing in the LC part because they are insufficiently damped. Fortunately the final RC section attenuates this. What are they for?

I am playing with ideas for SE amps. To see what is possible and how small you can get the PSU and keep the quality of supply. The performance seems to be similar for less parts count with the first.

Do you have any ideas that may improve the circuit?

Regards
M. Gregg

DF96 · 12th April 2011, 05:39 PM

I've just realised that your main output is taken from before the final RC, so PSU ringing could be a problem. You could check by changing the load CCS to a resistor, which will both damp the ringing and better model a typical amplifier load. Make sure any ringing is either well-damped or well below the audio range (and clear of arm-cartridge resonance if you use LP).

ryuji · 12th April 2011, 06:16 PM

why you using two inductors. i plugged it in real fast with same rectifier/load/resistance and get clean +-31mV on first tap, and +-62uV on second tap. time to play with the bridge rectifier one. heres pictures of my results.

M Gregg · 07:06 PM

Quote:

Originally Posted by ryuji

why you using two inductors. i plugged it in real fast with same rectifier/load/resistance and get clean +-31mV on first tap, and +-62uV on second tap. time to play with the bridge rectifier one. heres pictures of my results.

It's interesting if I model the bridge with the same values you use, it looks clean like your pictures. However it sounds realy bad! If I use the PSU as shown in the first post with diode bridge the PSU is silent, sound is very clear!(I mean no noise ear touching the speaker). With values that show the "good" trace as in your pictures it has 100hz ripple sound at the speaker! I mean loud! The difference as I see it is the level of noise on the first cap!

This is what I find interesting!

Regards
M. Gregg

ryuji · 12th April 2011, 07:27 PM

large input capacitors requires more current to drain. my intuition is that this is the reason you see the difference. If the psu sounds good tho, use it?

tomchr · 13th April 2011, 12:14 AM

What's the idea behind the current tap?

Glens · 18th April 2011, 06:58 AM

Dear M Gregg,
It has been a while since I used PSUD, but the values I quote have worked in many amps.
The PSUD models of both power supplies fall a bit short of the ideal.
The solid state PSU relies upon too much first cap.
The choke is too small.
Alter the 2H and 600 uF in PSUD and watch the ringing frequency on start up.
A 2 H choke has insufficient Z at 100Hz, 4H would be a minimum.
Use NTC thermistors in the mains primary to limit inrush current in solid state PSU's.

The Valve PSU model makes no allowance for the required series resistors in the HT trans legs.
Refer to the valve rectifier manufacturers datasheet for suitable values.
Omitting these allows the trans to ring like a bell and the rectifier may pop it's cork.
Don't believe the myth that none are required at low voltages.
To model this in PSUD, add the required values to the trans sec DCR. Eg.. 2 x 50R + 31 R DCR = 131 R.

I assume your design is dual mono.. If you set the PSUD constant current drain to step by + 20mA after 1 second and then - 40mA 1 sec later, you will get an idea of how the PSU will respond to transients. This is a rough approximation as the only way to test it is to build it....but it will show the difference between bad and better..

If you haven't already bought the parts for a CLCLC valve PSU, try modeling a mains trans with about 130 R combined DCR and series resistors, then a valve rect followed by a 47uF first cap followed by 4H choke, then 100uF and 10H then 200uF as a place to start. Fudge the input voltage as required to get the required output voltage.

Use a good choke with low DCR and ample max current rating. Add a tiny bit of series resistance to the legs of the caps to tame any ringing. (Model this by increasing the ESR of the first 2 caps by 2 -10 R.) Don't worry about fancy non inductive resistors, just make sure the max voltage and dissipation aren't exceeded.

For the low voltages you require, a choke input may be possible, try 10H first, 47uF Cap, it will cost more than a CLCLC design, unless you have the parts.
The peak current rating of the first choke in a LC filter must be at least 150% of quiescent, if I remember correctly.

A very good cheap and cheerful solution is to replace the second choke in the CLCLC design by any of the millions of Maida type regs, use an IRF 840 and LM 350 K (T-03 Steel package) and thoroughly test each PSU at max output using a dummy load for 30 minutes and then turn it on and off 10 times before risking the amp. Make sure to drop at least 30 Volts to avoid regulator drop out and the right heatsink is much bigger than you think, calculate the max dissipation and use 1 watt 15 V zener diodes in the gate circuit. Gate stopper resistors of 1K to 10K are required, I have used 47 K for certain applications. If the Zener or gate resistors go out to lunch, the full voltage will instantly appear at the output.....

If you aren't experienced with HV regs, don't try them as a small mistake may be disastrous. The cost of using a Maida reg is much less than buying a good second choke. Unless your amp is exceptional, a well built regulated supply will certainly not be the weak link..
The parts to make each reg cost less than $50 Australian, so don't let anyone shaft you.
Buy quality parts, as the cheap stuff will cost a fortune when it expires and takes expensive stuff with it..
I am unsure what the policy here is on naming brands etc, but the Swedish manufacturer of inductors whose part numbers start with LL.. always gets my money.

Although you can't model it in PSUD, using a hybrid bridge where 2 of the 4 SS diodes are replaced by valve rectifiers works very well, as the forward drop helps tame the noise and the valve warm-up time constant quells the inrush. Another advantage is the cost savings of no centre tap on the trans.

PSUD is a great way to get an idea of the basics. If you have a CRO and HV test gear, you can get a much better idea of the PSU's performance.

If you don't have a copy, buy Morgan Jones' Valve Amplifiers for the basics.

I hope this helps.

Regards,
Glens

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Quieting a Switching PSU? or easier to replace it with a linear PSU?	at77	Power Supplies	23	18th March 2010 05:46 AM
Aleph PSU design thoughts	Buhl	Pass Labs	3	13th January 2009 03:09 PM
Choke loaded PSU versus VR tube regulated PSU?	G	Tubes / Valves	37	20th March 2008 10:45 PM
Thoughts on this datasheet PSU ? LH0041	redrabbit	Power Supplies	2	3rd April 2007 01:26 AM
PSU questions and thoughts	CJ900RR	Chip Amps	1	3rd April 2006 04:30 PM

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

12th April 2011, 08:12 AM	#2
mastodon diyAudio Member Join Date: Jan 2011	The first one looks like the more affordable option, although you'll have to make sure the diode bridge doesn't get fried due to the high capacitance following it. I'd personally prefer this one vs. the tube rectified, double-choke power supply because the latter is relatively expensive and occupies a lot more space.

12th April 2011, 09:47 AM	#3
DF96 diyAudio Member Join Date: May 2007	Both suffer from ringing in the LC part because they are insufficiently damped. Fortunately the final RC section attenuates this. What are they for?

12th April 2011, 05:39 PM	#5
DF96 diyAudio Member Join Date: May 2007	I've just realised that your main output is taken from before the final RC, so PSU ringing could be a problem. You could check by changing the load CCS to a resistor, which will both damp the ringing and better model a typical amplifier load. Make sure any ringing is either well-damped or well below the audio range (and clear of arm-cartridge resonance if you use LP).

12th April 2011, 06:16 PM	#6
ryuji diyAudio Member Join Date: Feb 2011 Location: massachusetts	why you using two inductors. i plugged it in real fast with same rectifier/load/resistance and get clean +-31mV on first tap, and +-62uV on second tap. time to play with the bridge rectifier one. heres pictures of my results.

12th April 2011, 07:27 PM	#8
ryuji diyAudio Member Join Date: Feb 2011 Location: massachusetts	large input capacitors requires more current to drain. my intuition is that this is the reason you see the difference. If the psu sounds good tho, use it?

13th April 2011, 12:14 AM	#9
tomchr diyAudio Member Join Date: Feb 2009 Location: Greater Seattle Area	What's the idea behind the current tap?

18th April 2011, 06:58 AM	#10
Glens diyAudio Member Join Date: Jan 2008	Dear M Gregg, It has been a while since I used PSUD, but the values I quote have worked in many amps. The PSUD models of both power supplies fall a bit short of the ideal. The solid state PSU relies upon too much first cap. The choke is too small. Alter the 2H and 600 uF in PSUD and watch the ringing frequency on start up. A 2 H choke has insufficient Z at 100Hz, 4H would be a minimum. Use NTC thermistors in the mains primary to limit inrush current in solid state PSU's. The Valve PSU model makes no allowance for the required series resistors in the HT trans legs. Refer to the valve rectifier manufacturers datasheet for suitable values. Omitting these allows the trans to ring like a bell and the rectifier may pop it's cork. Don't believe the myth that none are required at low voltages. To model this in PSUD, add the required values to the trans sec DCR. Eg.. 2 x 50R + 31 R DCR = 131 R. I assume your design is dual mono.. If you set the PSUD constant current drain to step by + 20mA after 1 second and then - 40mA 1 sec later, you will get an idea of how the PSU will respond to transients. This is a rough approximation as the only way to test it is to build it....but it will show the difference between bad and better.. If you haven't already bought the parts for a CLCLC valve PSU, try modeling a mains trans with about 130 R combined DCR and series resistors, then a valve rect followed by a 47uF first cap followed by 4H choke, then 100uF and 10H then 200uF as a place to start. Fudge the input voltage as required to get the required output voltage. Use a good choke with low DCR and ample max current rating. Add a tiny bit of series resistance to the legs of the caps to tame any ringing. (Model this by increasing the ESR of the first 2 caps by 2 -10 R.) Don't worry about fancy non inductive resistors, just make sure the max voltage and dissipation aren't exceeded. For the low voltages you require, a choke input may be possible, try 10H first, 47uF Cap, it will cost more than a CLCLC design, unless you have the parts. The peak current rating of the first choke in a LC filter must be at least 150% of quiescent, if I remember correctly. A very good cheap and cheerful solution is to replace the second choke in the CLCLC design by any of the millions of Maida type regs, use an IRF 840 and LM 350 K (T-03 Steel package) and thoroughly test each PSU at max output using a dummy load for 30 minutes and then turn it on and off 10 times before risking the amp. Make sure to drop at least 30 Volts to avoid regulator drop out and the right heatsink is much bigger than you think, calculate the max dissipation and use 1 watt 15 V zener diodes in the gate circuit. Gate stopper resistors of 1K to 10K are required, I have used 47 K for certain applications. If the Zener or gate resistors go out to lunch, the full voltage will instantly appear at the output..... If you aren't experienced with HV regs, don't try them as a small mistake may be disastrous. The cost of using a Maida reg is much less than buying a good second choke. Unless your amp is exceptional, a well built regulated supply will certainly not be the weak link.. The parts to make each reg cost less than $50 Australian, so don't let anyone shaft you. Buy quality parts, as the cheap stuff will cost a fortune when it expires and takes expensive stuff with it.. I am unsure what the policy here is on naming brands etc, but the Swedish manufacturer of inductors whose part numbers start with LL.. always gets my money. Although you can't model it in PSUD, using a hybrid bridge where 2 of the 4 SS diodes are replaced by valve rectifiers works very well, as the forward drop helps tame the noise and the valve warm-up time constant quells the inrush. Another advantage is the cost savings of no centre tap on the trans. PSUD is a great way to get an idea of the basics. If you have a CRO and HV test gear, you can get a much better idea of the PSU's performance. If you don't have a copy, buy Morgan Jones' Valve Amplifiers for the basics. I hope this helps. Regards, Glens