How to make an audio console's power supply better?

[Dead_Nid]

Hello everybody! This is my first post. I've been around electronics-lab for the last 2 years, occasionally reading some interesting threads but now it's time I join in. I am not an electronics guru so I am more of asking for help than giving but I will try my best.

My point is on audio console's (audio mixing desk) power supplies, issues involved and ideas for improvements. I don't know if there has been relevant discussion on this subject in the past. If it has then please provide the link(s).

So...I have a Soundcraft Sapphyre (any audio freak in here will know about). It is a rather medium to large desk with 34channels. The console is being supplied by 2 rackmounted PSUs. These provide the various needed voltage rails (+17V,-17V,24V,+7,5V,-7,5V,48V). Each PSU has a toroidal transformer, some rectifying diodes,some big smoothing caps,some filtering caps and some positive adjustable regulators.

The regulators that Soundcraft put some 10 years ago when the unit was manufactured is Linear Technology LT1083CK (metallic K package). These regs are now obsolete and Soundcraft provides LM338K in their place. First of all I am suspecting that here we have an unpleasant compromise. I mean that even though I replaced all regs with LM338Ks and the console is working fine, the LT1083CKs must have certain qualities that make them better for audio applications. Does anyone have a clue?

This is the page for LT1083CK:
http://www.linear.com/pc/productDetail.do?navId=H0,C1,C1003,C1040,C1055,P1282

Do you know of any other "super duper fantastic" positive adjustable regulators of other companies that can provide voltages in the 7,5-24 volt range (in K package preferably)?

And a second thing that matches my subject better. My PSUs are 10-14 years old. Some caps had dried out, they made random "bzzz" sounds on the speakers, so I replaced them. But I want to ask in general: Would I have any improvement by replacing all components on the pcb? I mean all caps, diodes, transistors, even resistors. The PSUs had Jamicon caps from factory. Does anyone know if these are any good? What caps would you recommend no matter the price?

And the final most critical question: If you replace components that are equal to operation but have higher tolerances, does that matter? For example: The manufacturer has installed 10.000uF at 35V. You replace them with a 10.000uF at 40V or 63V. Is that OK or should you go for as low as it can get? Same thing with diodes. Of course I am talking about an audio PSU, not any application.

Sorry for the long post. I feel we can extend this thread and share interesting thoughts about this subject that has its own special needs (at least for audiophiles!). Thank you all! 
I can provide the schematics of my psus if needed.

 

[ante1]

Hi Nid, (I feel it’s better than calling someone Dead) ;D

Welcome to our forum.

I like to see the schematic first, and then it is easier to suggest which parts to check and which should be changed. 
It’s ok to use caps with higher voltage rating (definitely not lower rated) and yes, Jamicon is good. The LM338K is not a bad substitute as long as its current rating is within the limits (we don’t know enough about the PSU yet). 8)

 

[Dead_Nid]

what the hell!  ??? sorry I posted my reply as a new topic. I am sure I pressed REPLY!!! anyway,sorry  :'( :'( :'(, if it was my fault it won't happen again...check the topic "schematics for my Soundcraft's PSUs" and we can carry on in this thread  :'(

 

[ante1]

Hi Nid,

Well, shit happens and it’s not a big deal. ;D
The output from the regulators are not suppose to change their voltage when loaded, at least not if you measure right at the output. If you measure at the other end of the wires (load end) the voltage might change some under load due to the resistance of the wires. I would adjust with medium load while measuring at the regulators output, if there is a big difference in voltage between psu end and load end of the wiring the circuit might be improved with heavier gauge wires and if possible shorter wires. Do not re-route the power wires to cross any circuit boards though, since this can result in interference. I don’t think you should change the regulators but the caps and perhaps the pots too. However you might put caps on the rectifier diodes and on the diodes in the bridge rectifiers, use 10 - 22nF caps across each rectifier diode. I am not sure how much difference it makes but I would try it.  8)
 

 

[Dead_Nid]

Ante, why do you consider the other end of the wires as the "load end" when there is a whole mixing desk with plenty of circuitry just beyond? This is the load I was refering to and you are right we must include the wires coming from the PSUs too.I've always been measuring right at the regs outputs when the console was on. If it is off then I measure about 0.3V higher in each rail, in other words the voltage drop is about 0.3V and I think this is natural. Now what would a medium load be in this case?

Forgot to mention that one of the old regulators failed after 10 years or so in the previous owner's studio.That's why I got scared and decided to change all of them just in case. But I am always anxious that the old ones (LT1083CK) are better in terms of quality. I'm talking on specs (as far as I can understand them) and not that I've heard the difference in the sound or something. The LT1083CK is rated at 7,5A whereas the LM338K at 5A. Also the LT1083CK was twice the price of the LM338K. Now I can only get them at 60 euros each from a limited stock but that's not something I can afford (6 of them = 360 euro just for regulators!).Of course I still have 5 of the old ones but I am not so willing to use them.

So you suggest changing caps and pots. What caps company would you recommend and why? Is there anything special one should consider when buying pots?

Is there any online store (in Europe preferably) where I can find and order some premium quality electronic components?

Thanx,best regards.

 

[ante1]

Hi Nid,

It might be the wrong expression but I consider the consumer end of the wires as the “load end”. A medium load could be what you normally have switched on when using the equipment; only you (the user) would know that. I don’t know the layout inside but do you think it is running cool or have you noticed any “hotspots”? To prevent any regulators from destruction make sure they have large enough heatsinks and that they are securely fitted. On older equipment I usually remove the semiconductors from the heatsinks and clean the surfaces carefully and then put them back with fresh grease since this improves the cooling. What is the maximum current drawn (full load) through the LM338K regulator and at which voltage across? If you buy low impedance caps for 105C they are usually good quality. I know Panasonic and Jamicon works well but there are a dozen other brands that might be just as good. If you can find your country’s flag here: http://www.farnell.com/ it is a good source for high quality parts.  ;D

 

[Decade]

Hi!

I think the only modifications that has to be done is to replace the electrolytic caps and pots, these two components usually goes bad in PSU,s  I would go for all suggestions Ante wrote they are all truly good tech advices but the LM317K,s get tired once in a while also and they are cheap so to replace those doesn't hurt that much.
For the phantom power (+48) volts there is no need to worry, not much current are drawn from that regulator.
A voltage drop of 300mv isn't much but could be improved by heavier gauge wire as Ante said, when it comes to adjust the regulators I do like this, for constant load I measure right at the load but for a load that changes I adjust it somewere in middle of the span.
Since they use LM317K as regulators the designer has taken small loads in consideration.
Even if you use all channels I don't think it will put that much load on the LM317K,s some of these just feed the op amps, in one of a mixer I repaired I had to replace som of the filter caps in 3 or 4 chanels cause it was making a squeeling sound ;D
Why not put two quiet fans behind the rear of both PSU,s it will transport some of the heat.

 

[Dead_Nid]

Helloooo!
Forgot to ask about this:

However you might put caps on the rectifier diodes and on the diodes in the bridge rectifiers, use 10 - 22nF caps across each rectifier diode. I am not sure how much difference it makes but I would try it.

Is that some kind of extra filter that you suggest? By across you mean one cap in parallel with each diode? What kind of cap?

It might be the wrong expression but I consider the consumer end of the wires as the “load end”. A medium load could be what you normally have switched on when using the equipment; only you (the user) would know that. I don’t know the layout inside but do you think it is running cool or have you noticed any “hotspots”? To prevent any regulators from destruction make sure they have large enough heatsinks and that they are securely fitted. On older equipment I usually remove the semiconductors from the heatsinks and clean the surfaces carefully and then put them back with fresh grease since this improves the cooling.

The only hotspot is this. The PSU that has twice the +-17V rails was stupidly set to 220V input voltage. That made it work very hot all these years from approx. 1990 in the old studio without anyone noticing. Until one regulator finally died and then I bought the mixer in a really nice price

 

[ante1]

Hi Nid,

Yes, one cap in parallel with each diode and I use metallised polyester caps for this. Perhaps there are some better alternative but I know it works.
I think you should consider Decade’s suggestion to put a couple of small quiet fans in the PSUs. In combination with new fresh 105C caps and some new pots this part of the equipment should run well for a very long time. Regarding your thoughts on the LM338K regulator; as long as you are well within the specs (you should measure the current) you can relax. If still worried you can get the TO3P type (LT1083CP) of the LT1083 regulator for less the 20€, it will probably (almost) fit the heatsink too. The PCBs; maybe you can make new PCBs and completely rebuild the “powerplant”? Your second great rule “audio signal quality”; well, there might be hundreds of caps and pots to change in this department. Even screened wires and some connectors and switches could be in need of service, I don’t know the state of your machinery! From Decade’s post I assume he’s had his hands in a similar piece of equipment and might have some good advice for this part. ;D

 

[Dead_Nid]

Ante could you explain the benefit from the modification you suggest (caps across rectifier diodes)?

I am of course aware that audio signal quality is greatly depending on the state of the console's internal circuits and that means hundreds of components like you said. But let's forget that and concentrate on the power supply that is of critical importance audio-wise in my opinion.It MUST provide very "clean" current. I think that is the quality I was refering to.

Let me tell you another story now. The official spare parts manager in UK told me that the company is using LM338Ks in place of LT1083CKs because these are obsolete. That is definitely OFFICIAL. And so I did replace them. A few days later a

 

[ante1]

Hi Nid,

Ok, lets stay on the power plant then. ;D  I will make it easy for myself and tell you that the caps will reduce the noise produced in the diodes. The diodes make noise when constantly going from conducting to non-conducting state, higher load more noise. If you have access to an oscilloscope you could follow the improvements step by step. If you like to go real deep into this how about making new PCBs as I suggested before, it would allow you to optimize the layout of the components and perhaps add a few parts if needed. I don’t know the quality of the layout you’ve got today and nothing about the current or the voltage drop the regulators handles but I think some small improvements can be done, but ... is it worth the costs and trouble? As I mentioned the LT1083CP is less than 20€ if you prefer this solution. The oscillation the tech talked about is a bit like acoustic feedback which I am sure you know, but it can be very destructive in a regulator circuit. However I can’t see why the LM338 would be more likely to oscillate than the LT1083 in this circuit, they are not that different. The LT1083 has a very low dropout voltage, which I do not think necessary in this circuit. There are a few unknown (X) in the equation still. 8)

 

[Dead_Nid]

Thank you Ante, this is really nice info.

I don't think I will go that deep and make new PCBs. The suggested maintenance and modifications should make things just fine.

Can you provide a source (internet link,book) of documentation on the modification you suggest? And any source that I can study and learn more about power supplies as mine.

Finally how can I build this multi-amp-meter that I mentioned before? I mean an ampmeter for each rail.

Best regards,Nid.

 

[ante1]

Hi Nid,

There are many places to study linear power supply techniques, on the web and in books. You can try Google or if you got the ARRL handbook (your library might) there is one good chapter (11 I think it is) about PSUs. I am sure there are hundreds more, if I find an e-book on this subject I’ll post it.
Measuring the current on the outputs of the PSUs has some drawbacks. The shunt needed to measure current will increase the voltage-drop of which you already got some in the wires. The extra wiring (to the meters) could (worst case) pick up interference form other circuits and wiring in the equipment which then might find it’s way to the amplifier circuits (you don’t want that). It is of course possible to work around these drawbacks but you need good knowledge of the equipment as well as general electronics skills.

BTW. I am posting a couple of audio books later today in the Electronic Resources.

 

[ante1]

Hi Nid,

This might be interesting for you!?

[attachment deleted by admin]

 

[Dead_Nid]

A million thanx Ante, this is great!!!  :-*
On to work now.

Best regards! Nid.

 

[Dead_Nid]

hope someone checks this question out:

Is it natural for a polarised aluminum electrolytic capacitor to get hot? What can the reasons for that be? I am not talking about small ones (let's say at least 4700uF @ 16V).

Thanx!

 

[audioguru2]

Hi Dead,
A rectifier's main filter cap can get hot if its ripple current is too high for its ESR (equivalent series resistance). P= I squared R. Big capacitors are rated for max ripple current which occurs due to the cap charging from the rectifier then discharging into the load and with a high load current then there is also a high ripple current.

Replace your hot capacitors with ones with a high ripple current rating and a low ESR.

 

[Dead_Nid]

Hi! In the rail were caps get hot I measured 21.7VDC and 0.04VAC from the rectifiers' output whereas in other rails (where caps don't get hot) the AC voltage was max 0.02VAC. These 40millivolts can't be the cause for heat in the caps, can they? Caps are within specs. Old caps were also getting hot in the particular rail. I don't think that there should be a higher current drawn from that rail than from the others (with same outputs) but I am not sure about that. It has same design but different components though. Should I replace the diodes? (current diodes are "standard recovery")

Thanx, Nid.

 

[ante1]

Hi Nid,

I would say the cap is a goner! As Audioguru says; “Replace your hot capacitors with ones with a high ripple current rating and a low ESR.”.