SSlmixed.com Forum
Tech discussion => 4K,6K,8K Series => Topic started by: PMcC on November 12, 2013, 01:20:01 AM
-
What a great forum...
My aging 4K power supply has played up once too often now and I have decided to replace it with a 5K supply. Just waiting on the arrival of some cable so I can bring separate DC feeds to each batch of eight channels and then I should get to finish mixing the last song of a new album.
Just wondering if anyone is in a similar situation and needs a bit of 'step through' to get back into operation. There is a bit of reshuffling the logic lines to make the thing work but nothing too mad. I've powered up 16 channels plus the centre section so far and all seems good.
-
What a great forum...
My aging 4K power supply has played up once too often now and I have decided to replace it with a 5K supply. Just waiting on the arrival of some cable so I can bring separate DC feeds to each batch of eight channels and then I should get to finish mixing the last song of a new album.
Just wondering if anyone is in a similar situation and needs a bit of 'step through' to get back into operation. There is a bit of reshuffling the logic lines to make the thing work but nothing too mad. I've powered up 16 channels plus the centre section so far and all seems good.
Very interesting,
in my short experience with 5K I thought it was a bit too over-engineered and complex to deal with (too many protections and so on) but it would be VERY good to see some pictures and/or a diagram of how you have connected it.
Mattia.
-
Hi Mattia, the 5K supply is superbly engineered and beautifully partitioned into brute force Raw DC, individual 5A regulator cards and current monitoring sections.
I'll be offering up a full appraisal of the merging, with visuals as appropriate, paying particular attention to the impact this might have on the overall sound of the console, general reliability and serviceability.
All regulated outputs pass the load current through high current series diodes allowing paralleling of same voltage outputs with the attendant doubling, tripling etc of output current depending on how many outputs are paralleled.
Perhaps of general interest, the centre section alone consumes approximately 75 watts (minimal lamps on), the positive Audio rail draws 1.1A DC under same conditions, well within the 5A capacity of a single regulator card, mind you I've long since extracted the rear left and right mix bus compressor cards.
Eight channels consumes some 120watts, again with bare minimum LEDs and lamps on, and 1.26A of currents the audio + line. It's sobering to see how current consumption ramps up significantly with the dynamics sections gate LEDs in full flight.
For now back to finishing off the cabling...some heat shrink and flexible braided sleeving arrived today.
Back soon...
-
Under preliminary tests...
-
Well done!
:)
Mattia.
-
Thanks for that Mattia, just a quick update...
The wiring looms that connect from the 5K regulators to the 4K console are now built and ready for attachment. I have decided to avoid using the BICC 19 way connectors, preferring a more direct connection at both ends of the looms, I know I will have to sacrifice a degree of interconnection convenience by adapting this approach but how often do we make the disconnect with the BICC looms anyway? When installed I'll take some photos to expose the detailing.
As a related aside, when the original PSU failed it took out a rip of group amplifier summing ICs (LT 1028), definitely not a 50p part :-/ plus the 10R local decoupling resistor on the positive rail on that card.
Also even though there will be five separate looms connecting PSU to Console, I have also decided to parallel the individual regulator cards together, a move that will lower the overall output impedance of the PSU as seen by any circuit connected to it...this is a good thing.
I should have more to report in a day or two.
Pat
-
Stop the press...just finished wiring up the SSL 5000 series power supply to the SSL 4032 E series console, five separate looms in total so some reconfiguration needed at the console end (details to follow).
Powered up each section, one after the other with out smoke, always good and it has been running for an hour at this point with out any funnies so here's hoping.
Next up is a full on audition to evaluate the audio quality.
Pat
-
Back to life...
-
Back to life...
Well done!!!
can't wait to see more pictures and hear some audio tests!
Mattia.
-
I would consider putting a current meter on each module, its the only way to truly
make sure everybody shares the load. Good work!
-
Hi Xmax, that's a good idea. SSL describe a set up procedure based on voltages e.g. The Audio+ supply shoud be 19V so one regulator is set to this and all other Audio+ regulator voltages are differentially measured with reference to this value.
They specify a maximum deviation of 50mV from the ref. level but no mention of current hogging or sharing for that matter. I do like your idea of checking individual card currents all the same, no point in having a slacker amongst the ranks.
Pat
-
current is the actual thing you care about. Setting to the 50mV spec *probably* does the job just fine. But measuring it directly does so with more assurance.
You'll need a good DC probe, probably hall effect. Repeatability is what matters, not accuracy, so you may be able to save a couple dollars on that spec.
And don't forget about thermal tracking. If the two supplies don't track well then the drift over time may cause mayhem. Been there, done that.
this goes for all shared supplies.
Alan
-
Hi Alan, thanks for the input.
I may be wrong but my understanding is all individual regulator cards have power diodes in series with the output pass current of each voltage rail. The Audio+ rail for each card is say 19V as measured at the anode, the voltage at the cathode will be somewhat less than this and a function of actual load current.
If the cathodes are all commoned and one regulator card tries to deliver more current than all the others then it's series diode will have a larger volt drop than all the others. This is of course not possible as the other regulator outputs would have to drop in order to keep their series diode voltages at the lower current level.
Negative feedback holds the regulator output voltages at a constant value, this is the anode side of the series output diode. The cathode side is outside the feedback loop and therefore free to change with current load. Since the cathodes are all commoned they are all assuming the same voltage. If one regulator tries to runaway current wise it's series diode will try to drop a greater voltage than the other regulator series diodes implying all other regulator 'cathode side' regulator output voltages will drop. This is course impossible given the regulator outputs are held constant by negative feedback.
So it seems to me then that no runaway is actually possible. I may be off beam here but my view is the SSL engineers are/were extremely competent designers and why would they omit the whole current sharing issue from the set up manual if it was even remotely possible?
Having said all that I plan to measure the current from each card just to see his close they all are.
Pat
-
I assume you are talking about the 5K supply, which I do not know. I was mostly refering to the 4K/6K supply, which does not include steering diodes in the supply, they did have them in the combiner box, about all that is in there. It does not prevent the supplies from blowing up.
In the 4K what can happen (in rare cases) is that the supplies do not track and one ends up supplying the current as they deviate in performance. This them goes into thermal overload and the 4K supply when out of whack can do all sort of nasty things to itself, (not the board, that is protected). Mostly they just blow up and shut down. Not fun. Or at least that is what I surmise from the postmortem. Which is why adding the fuses is a very good idea, can save a lot of money and time.
I just re read your previous post, the problem between the two systems is that here, I assume that the one supply goes off line and the other takes over and then has a current overload situation, because there are only two supplies. This causes the crowbar to throw and kills the one supply. You are dealing with multiple supplies which together can sustain the loss of one supply, the old 4K situation couldn't, depending on the size of the board. A much better scenario for survival.
Alan
-
Hi Alan, that's a colourful raft of experiences you have had right there with the 4K supplies.
The move to the 5K supply, in my situation, is out of neccessity but it sounds like it might actually be for the better!
Pat
-
agreed
:-)
the standard 4K supply works quite well, when working correctly. Its failure modes are another story. That is where the newer supplies are way better from SSL.
-
I often see cases were 4/6/8k and 9000J's power supplies are not sharing the load properly
the overworked unit gets blown to bits and the other unit starts doing all the work until it smokes,
in the case of the 4/6/8k desks its trying to match 2 or 3 voltages of entire supplies with the 9k's
it's the switching modules in the power supplies, with the 9k's you just bump up the voltage
on the module that's "dead" the manual will only get you close, with the old monster 651's
you have to just keep letting them warm up then yank the beast and just try to rematch
all the voltages as close as possible, (they have a little "sponge" with all the diodes)
I have thrown out my back several times in the process, it takes hours, if anybody
says otherwise they just gave up and said close enough! The first thing I do when I walk
into a ssl room is make sure the power supplies are right, I have seen voltages off by 2-3 volts!
I have also had to parallel other supplies. I would stick a $10 amp meter on every module
and be done with it!
-
Hi Xmax, just wondering if the 5K console users out there (those that have survived the rush to the rack) have a history of PSU issues with current sharing. I think most 5Ks ended in radio/movie land so data might be a bit sparse.
The full pass current of each card's regulated outputs makes its way through 33mOhm power resistors on the anode side of the output diodes. I could monitor the voltages developed across these resistors to get some idea of current balance, might be a bit easier on the back.
I need to get a mix finished so time to put the partnership through its paces.
Back soon
Pat
-
Having run the console for a few extended sessions I can report all seems well on the 5K PSU front. Full details and pics to follow, post Album mix. One thing the 5K supply's lacks is the 263V DC for the plasma metering...my next point of focus.
Pat
-
Quick up date on PSU stability and nominal operating temperature...
Operational status: 32 channels plus centre section plus 'low voltage' metering circuits (no 250V DC for plasma yet!)
Period of usage: 12 hours (continuous)
Heat sink surface temperature: 34 degrees (2 ultra quiet 700rpm fans)
Location: Control Room
Importantly, the console sounds great.
The high voltage/power parts have arrived so I'm going to be building a stand alone 250V DC widow maker PSU for the plasma meters in the next few days.
Back soon...
Pat
-
Quick up date on PSU stability and nominal operating temperature...
Operational status: 32 channels plus centre section plus 'low voltage' metering circuits (no 250V DC for plasma yet!)
Period of usage: 12 hours (continuous)
Heat sink surface temperature: 34 degrees (2 ultra quiet 700rpm fans)
Location: Control Room
Importantly, the console sounds great.
The high voltage/power parts have arrived so I'm going to be building a stand alone 250V DC widow maker PSU for the plasma meters in the next few days.
Back soon...
Pat
THIS IS SO EXCITING!
CAN'T wait to see the plasma supply and some pictures of it all.
can you post some measurements, ripple, etc?Very curious !
one day maybe we'll have a community linear/modular supply for our beloved SSLs...
Mattia.
-
Hi Mattia, there seems to be a lot to commend the modular PSU approach, eggs in several baskets is one that immediately springs to mind. The bull dog end of the PSU, the unregulated bit that is, is still a single unit but it looks and feels nuke proof (again touch wood).
The 250VDC supply is based on a pair of parallel hi-volt Mosfets as per the original 4K design, hopefully it will hold up to job, should know very soon.
Regards
Pat
-
Direct connection of some of the DC power looms (DIY) at the 5K backplane on their way to the 4K, no BICC connectors!
Each individual loom carries Audio +(Red), Audio -(Blue), Audio 0V(Green), Logic +(White) and Logic -(Orange) cables. While the Green and Orange cables are connected using push-on 'spade' type terminals to the backplane, the other cables go to the output Contactors , it's just possible to see the Red, White and Blue cables disappear south.
Pat
-
...those relay driven Contactors (one for each bay on the console) that isolate the DC power in the event of faults.
-
...all five looms connected
-
Bam...action
-
2 years ago Funky Junk trashed 10 modules like this...
I have tried to sell them in any possible way, nobody wanted them...
what a waste..
M.
-
That is so unfortunate, I'm sure there are guys out there who would have given them a good home. And you know one of the great attributes of this forum is its strength of communication amongst the SSL user community, and beyond of course. Perhaps now one or two 5K PSU units will be saved from that trip to the dump...is this not recycling in a positive way? I think in the 50's Gibson had a somewhat similar problem with the original Les Paul guitar!
By the way, I assembled and tested the Hi Volt raw supply last night, it produced 326VDC (approx. 25mV ripple/off load. apart from a 22K bleed resistor). This supply will drive the 250VDC regulator which is almost finished and up for testing later today.
Pat
-
Well I survived the 250VDC build, test and implement.
This is interesting, the original SSL 250VDC supply, as it turns out, had drifted upwards to about 295VDC. This gave all the plasma meter units a soft focus appearance, a bit like out of focus. The offending part was a 220K carbon film resistor that helps define the regulated output voltage, it had gone way outside its tolerance band so the 50K trim pot had no chance of lowering the voltage back to 250V. There may be a few others out there that could benefit from a very simple fix, just replace the offending 220K resistor on the PSU mother board right beside the three trim pots associated with the 250V regulator card.
Anyway I built my own 250VDC supply and popped it into an unused power and chassis, great for heat sinks, mains connection etc and hooked it up to the Console after a bench test of course. Result...pin sharp plasm metering, right on focus, a thing of beauty. Not bad for a thirty year old board.
Oh yeah the 250V (plus small change) is holding rock solid and showing 25mV of ripple at full tilt.
-
No, I was just speaking in general with current sharing. I have found having current meters on each "module" is the only real way (or easiest) to equally balance the voltages. as far as your 250v circuit, I would love to see your version of the ssl design. Nice Work!
-
Hi Xmax, good to hear from you...I am sooooġ happy to have metering back, I know we all trust our ears more than anything else but just for that DAW playback line up, super.
When I get a clear minute I'll draw up a schematic and upload here with actual voltages at the key points. By the way if found 240V low wattage halogen lamps a great way of loading the hi volt supply for bench testing.
Back soon
Pat
-
Hi All,
Fact...32 plasma modules draw a maximum current of 164mA from the 250VDC supply.
How do I know this?
By taking voltage readings across the two 4.7R current sharing resistors that are in series with the two power Mosfets of the supply I built to replace the failed 4K unit, I have been able to calculate this current.
The voltages across these resistors are reassuringly similar, 384.89mV and 389.6mV. This equates to approximately 82mA in each resistor.
To draw the greatest current, I needed to light up every segment of every plasma module. This meant max'ing out all VCA faders with the metering switched to monitor VCA positions.
Clearly plasma metering is no power hungry monster, nice to know.
I've included a couple of pics, ok the HV supply build has a prototype look but it works well and being completely separate and independent of the main 5K supply, if the worst does happen, it can easily be dispatched to the 'bench' for fixing while the session can continue.
-
The voltage across one 4.7R current sharing resistor...all plasmas flat out.
-
Early doors for the assembly of the HV regulator power components...two International Rectifier high voltage Mosfets (TO220 package type) plus their associated 4R7 current sharing power resistors (ceramic), all four mounted on an 'L' extrusion bracket.
Pat