Console puzzler - SOLVED (sort of)

I ahve a DDA Dmr12 which has been great to me for the last 11 years, with the exception that the PSU was kind of delicate.


It seemed like it would need repair every couple years. A few years ago, it was recommended that I replace the PSU with 2 Power One F24-12A (24v 12A) supplies. and a Power One hb48-0.5a for phantom. I was able to source Condor branded equivalents.

The console ran on +/-18V and 48v phantom, so i needed to add a resistor to the voltage dividers on the 24v supplies to drop the outputs to 18v. The supplies can now be adjusted from approx 16v to 20v. I have them set at 17.8v each.

I have them connected correctly so that one +18v and the other is -18v from ground. That is one console supplies +18v and ground the other supply has it's +18v terminal connected to ground and It's ground is connected to the -18v terminal on the console connector.

My current (no pun intended) issue is that with all channel strips loaded, the console will not come up correctly.

Symptom is that the main L/R Vu meters peg and the bus meters also peg. No audio gets through.

The console has a 56 channel frame (32 input; 24 output in a split format) + 2 Aux strips with 4 aux on each, Plus the Master section.

If I remove 20 output channel strips, the console starts reliably. If I add in the strips I removed without powering off, the console continues to act correctly, passes audio, sounds clean, all functions work, etc.

If I add strips, but power down as each is added, then repower, I can add up to 2 strips, but it may take 3 or 4 tries to get the console to come up correctly.  Above that adding another strip causes the pegged meters no audio

I notice that the transformer on the +18v supply hums when the console is connected to the PSU, but is dead quiet with no load.

Would this indicate that that supply is not capable of providing anough current to meet the power on load?

How should I go about diagnosis?
I'm guessing that since the old supply only did 18V 10A that I need to be able to prove that the new 18v supplies can deliver that at around a 9-10A load. In fact at 24v they were rated at 12A, so they should be capable.

So a 2 ohm power resistor for a load would get me to 9a and if I'm not seeing 18v anymore from either supply, that I've found my culprit.

Is that true? I'd then suspect that I've got a pass transistor failure?

If the power transformer is humming noticeably when powering the console it could be the transformer is working too hard.
However, that in and of itself shouldn't prevent the console from powering up.

It could be that your power supplies are current limited and the inrush current when you first start up is exceeding the max current and tripping whatever limiting mechanism is in the PSU. With a heavy load and a bunch of capacitance, that inrush current when you first power up looks like a dead-short to the PSU.

If the constant current draw of the console is well under the max constant current rating of the PSU, you just need to figure a way to power the console in delayed stages so that the transient spike of inrush current isn't killing your PSU when starting up.
If your PSU can't handle the inrush current, if you keep trying to use it over time it will likely fail.

Using resistors to drop the voltage is a bad idea because regulation is greatly compromised, and a lot of power ends up being wasted as heat. Unfortunately there's no good way to use 24 volt supplies to provide 18 volt power. This may also be causing or worsening your start up problem.

> Using resistors to drop the voltage is a bad idea because regulation is greatly compromised

No, I read it as he tinkered the Reference Divider which sets the output voltage, not that he put a resistor in series with the output.

However a linear supply scaled for 24V and trimmed to 18V is wasting-off 6V more than the designer intended.

Efficiency (24+V models): 60%

If the raw rectifier gives 40V and there are no standby losses, then 24V/40V= 60% efficiency.

Trimmed to 18V we have 45% efficiency.

24V design wastes-off 40V-24V or 16V. 18V trim wastes-off 40V-18V= 22V.

At 12A load, 16V*12A 192 Watts dissipation. 18V wastes-off 264 Watts!

The 18V rating should be closer to 8.7A.

When loaded with stable load and run 20 minutes, how hot do the heatsinks get?

PRR, corrrect, the reference divider was altered.

The humming supply (the +18v unit) gets hotter than the negative. I can touch the heatsinks on either, but the negative runs noticably cooler. The cooler gets uncomfortable after I've been touching it for about 10 sec. The hotter one I can touch for about 4-5 seconds. Certainly neither is more more than 120 degrees Farenheit +/- a few degrees.

Neither transformer is above just warm. There are 6 2n3055 pass transistors on each supply. It seems like since each can dissipate 115 watts, that I should be OK. Or is the power dissipation 264 watts on each pass transistor.

It is not unusual for a console to draw more current on the positive rail than on the negative one. That why the positive is often much bigger...

I don't know the DMR12 but 10A for 56 channels seems really low to me. For instance, I've been working on an Amek that is drawing 6.8A with 24 channels.

It is not unusual either for bean counters at the factory to under spec the PSU for larger frames console. I.E.: same PSU for 24, 48, and 56 channels. Works fine with 24, hot with 48, and fails regularly with 56...

My point is your console was delivered with a 18V/10A PSU that was 'kind of delicate'. You replaced it with a new one which, as PRR explained, is capable of 8.7A... That's actually a downgrade.

You should add an Amp meter in series with the +rail to see what the actual current consumption is.

My personal rule is : I want the PSU to run cool a twice the needed max current. (That Amek ended with a 20A PSU...)

Axel

You might be running into op-amp/amplifier latch-up caused by the power supplies stabilizing at different rates.  Most bipolar supplies reference one rail to the other so that the voltages are symmetrical, if not always exactly what they're supposed to be.  The solution may be to find a power supply that does this, or find some way to make sure the absolute value of the difference of the voltages on the positive and negative supply during start-up is as low as possible.

Joe

You need to find out what your console is really drawing. Put an ammeter in each power supply leg.
Also I agree with the previous comments that you would be better with a tracking supply, and also
modifying a 24 volt supply down to 18 volts will get you into long term trouble, due to the added
dissipation. I would suggest what is happening is that the rails are not coming up symetrically, and possibly one supply is current limiting.

> There are 6 2n3055 pass transistors on each supply. It seems like since each can dissipate 115 watts

The 115W rating is for INFINITE heatsink. There are no infinite heatsinks.

The package can dissipate 1W-2W no heatsink.

Heatsinks come in all sizes. A silver dollar bolted to one ear gets you to 3W. An aluminum slab drilled with multiple high-velocity water passages could get you over 100W per device.

As designed: 192W dissipated in six devices is 32W per device which is not unreasonable for a large sink like that.

> I can touch the heatsinks

Heatsink tips run cooler than the mounting surface. Mounting surface runs cooler than the device. At the risk of 40 Volt shock, touch the 2N3055's lid. (Or cook till hot, power-down, then touch.)

If Axel is saying that wide consoles eat over 10A, then I suspect these modified 24V 12A power supplies are at or past the edge. Which agrees with your observations.

> Most bipolar supplies reference one rail to the other so that the voltages are symmetrical

They reference one to the other, but not vice-versa. If the + supply is master, and the - supply "tracks", then a collapse of the - supply still leaves the + supply running. And that sure would "peg the meters".

I'm thinking, after finding enough power, if there shoud be a "monitor". If either rail falls below 14V, BOTH supplies are forced-down, and a light indicates the reason.

Larry, I realize that two 24V 12A supplies is a major hit on the budget, and it is distressing that it isn't going to do the job.

Is there any chance the system will run on +/-15V? They used 18V because that is the rating on the chips, but 1.5db headroom is not worth big drama, and 15V supplies are a stock item, probably to 20A.

Is there any chance the board power can be split, 32 modules on one lump, 24+4+4 on another lump? Get expert opinion before considering this; I am not broadly experienced in such large consoles. My small-board hacker shivers at the thought of a board run on two supplies, however clearly you are putting modules in and out at-will.

It is not uncommon for larger desks to use multiple +/- Volts supplies for powering the audio circuits.  For example, the Amek 9098 "family" desks used a supply for every 16 channels (give or take).  Each of the MPS-15 supplies contained a pair of International Power 15 V linear modules (factory modified to provide 17.5 VDC) at around 12 Amps.  An additional factory mod was to provide an "inrush slowdown" so that the PSU modules wouldn't go into overcurrent limiting at start-up.

http://www.brianroth.com/projects/curtis/amekps.jpg

http://www.brianroth.com/pix/aaron/mps15/

Those PSU's are for these desks, respectively:

http://www.upstairsproductions.com/html/screen_shot_enlargement.html

The "split" version:

http://www.brianroth.com/pix/aaron/april2009

That same desk in its new home:

http://www.brianroth.com/pix/gilberto/IMG_0128.JPG

The PSUs:

http://www.brianroth.com/pix/gilberto/psu/

In the first pic, you can see there are two of the International Power 15-15 linear modules at the sides, and a 5V/25A (for logic) in the middle position.

Best,

Bri

I had a look at the DMR12 schematics: http://www.ddaconsoles.com/pdf/schematics/dmr12-schematics.pdf

On paper it looks like a better than average PSU. Fully discrete as far as the +/-18V rails goes, no exotic parts, good symmetry, seems to have some tracking...
Not sure the Condor would be really better...

But one thing bothers me: The OP said that "the old supply only did 18V 10A".
Then factory schematic shows two versions of the PSU, a 10A, and a 15A called HD for Heavy Duty.
Since the DMR12 came in two frames, 44 or 56 modules, one can assume than the larger 56 frame required the HD 15A PSU...
Could it be that you got the wrong PSU from the start?

If you still have the old PSU, it's easy to spot, the HD has 8 X 0.1 Ohms resistors on the collectors of the TO3 transistors. The smaller PSU only has 4.

Again, a couple of Amp meter in the +/-18 rails will tell...

Axel

I agree with the latch-up theory.  Either the opamps as suggested or the logic.  The +18 must be sagging, and the logic depends on it as well.  Have you looked at the supply voltages loaded vs downloaded?  This is one of those times when having a few cheapie DMM's is a help.  And the cheapies usually do 10A current where my Fluke is only about 2.

You should be able to do make specific current measurements for the modules and the master section.  I always plan for running replacement supplies at 60-70% of rating.

Amek on the other hand designed for about 95%!  Their Mozart console supplies ran so hot, and the replacement supplies were twice the capacity. 
Mike

Well, I got some readings. For some reason, my peak hold on my DMM didn't capture. so I've only got steady state data.

I unloaded the supply by 7 channel strips and was able to get it to start.

+rail steady state = 4.94a Add a channel = 5.04Add another = 5.11 indicating to me that each strip is drawing approx .10A

- rail steady State = 4.63A Add a channel = 4.72 indicating a delta of  .09A

Adding the additional 7 strips would seem to indicate constant current needed is 5.74A on the positive supply and 5.35A on the negative.

Each channel strip has an onboard regulator to derive 5V for the logic stuff.


I'll check the old supply to see if it's the HD or not. the + and - 18v rails were fused at 10A consistent with the faceplate markings.



EDIT:
Forgot to mention:

I measure current when the console fails to start.
+ Rail = 5.16A with 8 channels Pulled vs  4.94A when  started OK.
- Rail = 7.75A with 8 channels pulled vs  4.64A when started OK.

So just for the heck of it, I bought a Hong Kong special 18v 10A switching supply. It was only 30 USD.

A question for y'all. Why do linear supplies seem to be preferred? Assuming this meets claims, it would appear to be an ideal solution from both a price and size/weight perspective.

Larry Sheehan said:

I unloaded the supply by 7 channel strips and was able to get it to start.

+rail steady state = 4.94a Add a channel = 5.04Add another = 5.11 indicating to me that each strip is drawing approx .10A

- rail steady State = 4.63A Add a channel = 4.72 indicating a delta of  .09A

I measure current when the console fails to start.
+ Rail = 5.16A with 8 channels Pulled vs  4.94A when  started OK.
- Rail = 7.75A with 8 channels pulled vs  4.64A when started OK.

Click to expand...

If I read this correctly,
first set of measurement, both rails around 5A within 10%
second set of measurement, +2.5A on the negative rail!
Something is wrong...

According to the schematics the 'small' supply was fused at 10A, that would indicate you had the wrong one (Or at least not the best one)

The issue with switching supplies is that they are not always well filtered, especially when they are cheap...

Axel

Thanks, Axel. I'm flummoxed at this point. I don't even know what to look at next. I assume I still need to determine the cause of the no-start. Empirically, I think it's a given that the condors at 18v are current constrained.

I did notice another symptom of the no start which is that if no start, the solo and mute do not operate (Which I could be convinced means that it's the logic that is hung). The mute button drives a flip flop which in turn operates an IC switch. Solo drives the other half of the flip flop and operates the other half of the IC switch. The mute solo LEDs follow the flip flop.

You mentioned that the transformer of the +18V supply is making noise when it fails, but the negative rail shows more current in this configuration...
I suspect that the 5.16A on the positive rails is that low because the voltage is falling down...

You need to measure the current and voltage simultaneously. Ideally, if you have 4 cheapo meters, leave them wired until problem is solved.

Also, it's not clear from your op if you changed for the Condors a few years ago and it worked fine until now, or if you did the change recently and it never worked well?

Axel

Did you turn up the current limit trimmers to max ?... they are typically only half way up when you buy them.

(assuming that the Condor version is the same as the International Power and Power One versions it is next to the "Vadj" and is labelled "Ilim")

Colin
www.audiomaintenance.com

Justn switched to the Condors.
No I didn't change the current limiters. I'll try that today.

It appears that all those supplies are really made by SL Electronics and rebranded.

I'll hit RatShack and grab some extra meters non the way over there today.


Good ideas. Everyone.

WINNER! The current limiter trims were as shipped, about 5/8 up. I had to scrape some white glue off the adjuster screw, turnded them up to the max, plugged in all channel strips, and it starts reliable every time. Steady state load is 5.77A. Sounds good, all functions working.

I may still see what that switcher can actually do, since I bought it. If it can do 10A with low noise/ripple, I think I'll jump to it. That would allow use of my existing case, instead of kitbashing a  rackmount computer case I have. The Condors are too big to fit into the existing case with any breathing room.