GroupdDIY 511 Racks: round 5.5 available

[sahib]

I am glad Volker came to my help. I was actually replying with more detail and before I had a chance to hit the send button my pc froze. Bastard!.

Anyhow.

The power supply is designed to supply 50/50. But the entire rack being 11 modules and number 11 being odd this divides 6 modules on 16V and 5 modules on 24, or otherway round.

The test I carried out is under 1.6A current draw from each rail.

I still have not come across a clear standard on what the current consumption of each module is. But based on 150mA per module 11 module gives us just over 1.6A.

As seen from the test 16V supply is totally comfortable with 1.6A current draw. Therefore I can categorically say that if you run your entire rack loaded with 11 modules, and you do not exceed this current rating greatly then the power supply will handle it.

However, things are different on 24V side. At 1.6A the regulators are hitting over 90 degrees at tab. As mr normally nice but currently grumpy  due to a nasty flue also indicated the junction is probably around 100 degrees. Although still 25 degrees below the maximum operating tempperature, it is not low enough to run it under that load safely. But it will comfortably handle a total current draw of just over 1A. Which means 24v side can handle up to 7 modules based on 150mA current draw each.

If we sum up the discussion, you can;

1. Run 11 modules at 150mA each from 16V side.

2. Run 7 modules at 150mA each from 24V side.

 

[[silent:arts]]

example from a normal loaded rack

 

[jdbakker]

Well...

While I'm the first to admit that I prefer systems to be designed somewhat more robustly than the next man, I feel that dismissing it as

some kind of sport.

... is not entirely fair.

which module takes to full advantage of drawing the full specs at all 4 voltage rails at the same time?
I'm not aware of even one, and you would need to install 11 of those modules in the same rack to reach Mr Sahibs heat

That's not necessarily true.

I see no reason to assume that the two heatsinks in Sahib's pictures interact much. In that context, populating an 11-slot rack with modules that draw 130mA current from only the 24V-rails will overheat the PSU. Given Sahib's numbers, I expect that in a full rack on a hot day even the 16V-regulator will be close to or at thermal shutdown when feeding 11 modules that are at the limits set by the VPR Alliance specification for the 500-series.

(Of course, a PSU redesign or forced cooling as Sahib hinted may push the envelope a bit further)

in a "normal" installation, with eleven different modules, the PSU won't even get warm (as mentioned by some users)
most people won't use the 24V rails anyway. the specs are fine, at the real maximum it might get hot, that is normal.

So specify a maximum for the GDIY51x-compatible modules already.

(EDIT: I see Sahib's just done something like that. It would be useful to know if this is official GDIY51x-dogma, and if so it might be useful to put it on the 51x spec pages)

As a module designer, I would very much like to know how much current I'm allowed to draw, given that a user might put 11 of my modules in a single rack. It's all well and good that there are no modules approaching the VPR-limit right now, but that doesn't say very much.

Look at it this way: if I buy a rack+PSU from API and friends, I know that I can mix&match any and all VPR-compatible 500-series modules without worrying over current draw.

JDB.
[don't mean to keep overstating this, but IMHO in your last post you were downplaying the issue a bit too much]

 

[jdbakker]

However, things are different on 24V side. At 1.6A the regulators are hitting over 90 degrees at tab. As mr normally nice but currently grumpy  due to a nasty flue also indicated the junction is probably around 100 degrees.

Hotter than that, even.

If the reg was dropping its minimum 3V, the dissipation would be 3*1.6=4.8W. At the typical junction->case thermal resistance of 3K/W you'd end up with a junction that's 14.4 degrees hotter than the tab. That's ignoring the fact that the RMS voltage across the regulator will be higher than minimum, if only to account for reservoir cap ripple, and that the case temperature at the case->heatsink interface is likely to be hotter than the (accessible) spot you measured. And, as you noted, that's with the PSU sitting on your desk in a relatively cool ambient.

JD 'tickling the dragon' B.

 

[[silent:arts]]

Well...

While I'm the first to admit that I prefer systems to be designed somewhat more robustly than the next man, I feel that dismissing it as

some kind of sport.

... is not entirely fair.

sorry if there is any misunderstanding.
it was not negatively meant.

... So specify a maximum for the GDIY51x-compatible modules already.

(EDIT: I see Sahib's just done something like that. It would be useful to know if this is official GDIY51x-dogma, and if so it might be useful to put it on the 51x spec pages)

As a module designer, I would very much like to know how much current I'm allowed to draw, given that a user might put 11 of my modules in a single rack. It's all well and good that there are no modules approaching the VPR-limit right now, but that doesn't say very much.

I will change the official specs from "maximum power consumption per unit: depends on your PSU" to 150mA per rail.
at the end it is still "depends on your PSU", since everybody is free to choose the PSU on their own.
I can't / will not do any specs for 3rd party PSUs.

Look at it this way: if I buy a rack+PSU from API and friends, I know that I can mix&match any and all VPR-compatible 500-series modules without worrying over current draw.

now - the VPR specs are for the modules only, not the PSU.
there is no 3rd party case or PSU approved by API.

however, take care of your flue,
and don't stop posting your opinions about this PSU here.
your input and knowledge is very welcome.

but please keep in mind - while we want the best - this is a DIY unit, build by others.
we are neither Philips nor Harman with a finished commercial product.

 

[sahib]

Although it is going to sound biased because I am in a business relation with Volker, but I must agree with him that I do not see any reason why we would need to specify a maximum current draw per module. As long as we give the maximum current rating of the psu then it is up to the guys to decide how to load their rack.

On the psu help thread there is an example of a built psu which apparently did not heat up significantly under full eleven modules. With which, I must say I am exteremely surprised, though pretty happy. However, if the modules incorporate a simple jumper linking 16V and 24V rails together then 24V rails could also be set to 16V allowing more modules to run. Or otherway round.

 

[Spiritworks]

I'm not so confused any more , thanks.

 

[sahib]

Another test under more realistic conditions.

The 24V rails have been loaded with 25R resistors. Total current drawn 0.96A which equates to 6 modules on the basis of 150mA per module.

As you can see from the picture there is a fan blowing hot air onto the case. the temeperature around the case is over 35 degrees C. Feels like back in Istanbul.

One hour into the test the temperatures pretty much settled.

Regulator cases between 70.4C and 72.8 degrees. Tabs are between 80.0C and 82.4C

Hottest part on T section 41.6C

Hottest part on the heat sink 37.1C

Rectifiers at 53.0C

Gotta run to collect my son from the school. I'll run it longer tomorrow.

 

[jsteiger]

Come now Cemal. I wish you would just put a little effort into this.  ;D

You know I could not be more sarcastic.  8)

It's hard to express how impressed I am at what you are doing here! 

Much thanks from all of us! 

Cheers, Jeff

 

[Dubka]

Wonderful!

Now, when do we get to buy these beauties Cemal?  ;-)

 

[sahib]

Very soon. As I have just completed the final endurance test. We will work out the total kit price and start taking orders hopefully by the end of this week.

Now, this is the final test in real situation. The test took 3 hours in total as in ten minutes I have to go and collect my son from the school before he is chucked out onto the pavement. I am afraid this is modern life. So, anybody thinking of having kids think twice. No four times and give a year gap in between.

The PSU case is fully enclosed and mounted on the rack. There is a 1RU case on the top and bottom like in real world. I have put a sheet of mdf at the back, blocking the ventialtion as this normally be the case in the front when the rack is loaded with equipment in full. Fan is blowing hot air into the rack space.

The 24V rails have been loaded with 25R resistors. The current drawn from each rail is 0.96A which equates to 6 modules on the basis of 150mA per module.

Below readings were reached after the first 20 minutes and stayed pretty much constant.

The rack is metal and the average surface temperature is 35 degrees C.

Inside the rack, immediately under the 1RU case below the psu case the ambient temperature is 67.6 degrees C.
It varies but never below 67.0 C.

The surface temperature on the back panel of psu case is 50.6 degrees C.

Inside the psu case 48.9 degrees C.

(I have another xlr hole at the back of the psu case to insert the probe to take measurements. So, in this three hours the case has been fully closed and in position.)

Do not be misled by the VOM reading. It is not the phantom output. I am reading across both rails at 24V.

After the first  20 minutes when the above temperatures were reached the output voltage went down by 0.1V and this continued for the next 40 minutes eventually settling at 47.7V and remaining for 2 hours.

And we ain't got no fan fitted on the case yet.

 

[unlikekurt]

With the power supply kit looking like its still a ways away, having not been able to get anything more done with the rack kit than make sure that everything was in the package, and having some finished modules sitting on a bench for a few months without use, I'm thinking of putting my rack kit up for sale and biting the  bullet on an api box.  Anyone interested?

 

[Meathands]

Unlikekurt,

I might be interested in that, especially because I live in Harrisburg and might be able to pick it up.  I wasn't clear on whether PSU-only kits will be offered at some point.  Is that in the cards?

Thanks.

 

[sahib]

With the power supply kit looking like its still a ways away, having not been able to get anything more done with the rack kit than make sure that everything was in the package, and having some finished modules sitting on a bench for a few months without use, I'm thinking of putting my rack kit up for sale and biting the  bullet on an api box.  Anyone interested?

With all due respect but have a look at the photographs again. Does it look like unfinished? It is unfair to say that.


The pcb for the psu has been made available with the racks, so has the transformer.


The help thread http://www.groupdiy.com/index.php?topic=36874.0 has been up and running from day one with full sechematic and bill of materials. Many people already have built their own psus. 

However, what we intended is to eliminate the mechanical work for those who do not wish to do or do not have the means to do it.  If you are one of them then we now have arrived and don't panick.

I have completed the bill of materials and worked out the component costs. As it is impractical for me to go through every single supplier and find the product code, I based it on RS Components' one off prices. The only one from Farnell is the bridge rectifier which is a 25A in line type that is required. I have ordered few from to test but due to its high current rating I expect it to eliminate the need for a heat sink on the rectifiers.

So it will not make any difference to us whether you buy them from us or from RS or any other supplier. Therefore please bear with us for another couple of days.

Finally, thank you all again for your continuing support.

 

[unlikekurt]

Sahib:
I think I may have mispoken to a degree.  You guys are amazing, Cemal, Jeff, Volker, all amazing..  The project is amazing.  The PSU rack is certainly right around the bend and I'm sure it would have been available sooner if it weren't for the attention to detail and quest for perfection you all possess!  And the schematic and BOM was made available quite a long time ago, as well as the support of folks here.  I think I should have worded some of my post differently.  Fabricating the chassis was something I wasn't prepared to do and I haven't had the free time needed to get around to the rack (of course I thought i'd have oodles, but that never works out).  I apologize for any unintended criticism or insult, you guys are brilliant. 

 

[sahib]

Absolutely no apologies necessary and no offence taken. 

Indeed things have gone a bit slow but I can assure you that is due to many factors that is outside our control. In terms of metal work if you buy it off the shelf it is there and available but as soon as you start to say can you do this on, can you do that, then it starts to drag on. However, we are going down the route of buying off-the-shelf case and I will be doing the additional work such as mounting/ventilation holes, heatsinks etc. in-house. This will be much quicker. In fact I have already started working on five psu cases.

 

[mata_haze]

FANTASTIC!
don't forget to count me in for one of the kits!

Mattia.

 

[Meathands]

I poked around here for the answer, but just wanted to confirm: will the psu cases/hardware be available as a kit separate from the pcb?  Thanks...

 

[sahib]

PSU will be available including every single component, from the 2RU case right down to the screws and washers.  You will not need anything else.

However, I have grouped the components in different blocks such as;

1. Empty case with all mounting screw/ventilation holes.
2. All electrical connectors from mains switch to 7 pin XLR.
3. Heatsink kit.
4. Toroidal transformer.
5. Transformer mounting plinth.
6. Blank power supply board.
7. All the electronic components for the power supply board.

These will be priced individually allowing you pic and mix as well as buying them all.

Hopefully I will get the pricing for the heatsink profile on Monday or Tuesday and post the pricing by mid-week.

 

[[silent:arts]]

Cemal, great idea to split it up, since some people already have PSU PCBs and transformers.
good man as always ;D