lm317/337 too little current for 500 module???

[wkbdgeorge]

Hey all,
I've made a 317/337 based power supply to power a pair of 500 modules.  It has worked fine until I came across one brand of module that doesnt work.  The Vintage Design m581mk2 module is having issues with the supply.  Vintage design states they have a " power on delay to prevent PSU errors".  From what I see, they have a 7824 regulator that ups the 16v supplied to 24 for their DOA. 

Also, they have I think a pair of relays that turn off the voltage rails if it doesnt like what it sees.  So when you engage one of the lighted push switches on the front, it turns on and off and on and off and I hear the relay firing. Right now I'm just trying to power 1 module.  In the past I have powered a single great river two wide module, and pairs of avedis ma5, shadow hills mono gamas, and others without issue.

Soooo, is the lm317 too weak to power such a setup?  Does a fixed voltage regulator have more ooomph available?  Help!

 

[pucho812]

it all depends on current draw.

There are VPR compliant modules which adhere to API specs that include a max current draw per slot, I forget exactly off hand what the numbers are Then there are 500 modules that are not VPR compliant but fit the 500 form figure.  This can cause problems to anyone.  how much current can your PSU send out? The 317/337 combo can output a max of 1.5 amps with proper heat sinks installed. I would start looking at that

IIRC vpr specs each 500 module to draw a max of 110 miliamps current. So even if not a VPR compliant module, I highly doubt it's draw much more then that...

 

[wkbdgeorge]

I think the limiting factor in the design is probably the power transformer.  It has a current rating of 694mA.  But that should be plenty for one module!  There are 1n4007 diodes, lm317 and lm337, some caps.

Ive tested the module in an aphex brand rack and it does the same thing.  However, when I put it in an api lunchbox, its fine. 

Confuzzed.

 

[Brian Roth]

Measure the PSU rails with both a DMM and an oscilloscope when trying to run the errant module with your 317/337 supply.

Bri

 

[pucho812]

o.k. well what is your power supply schematic? And what aphex rack are you referring to?

 

[wkbdgeorge]

hmmm, so I was poking around with the dmm a bit more along with my scope from Brian's suggestion.  The module is an open design without a shroud so I was checking some voltages.  I found some weird stuff happening.  Let me know if this is normal.

The voltages at the input of the card are correct and are hitting the correct pins.  As I was poking around some more, I found that the voltage at its 7824 regulator to be NEGATIVE 16v both before and after the regulator.  Also, I was checking the rest of the audio input and output paths and found that at the output of the module, its putting out +16v on its ground pin. 

Could these modules just be faulty?  But why would they work in the API brand rack?  I checked my shadow hills mono gamas again and they dont exhibit this behavior.  Can lack of current make this voltage weirdness happen?

 

[wkbdgeorge]

Update:  I powered the module with an api brand rack and all of those weird symptoms go away. 

 

[wkbdgeorge]

So, I was checking out the API brand power supply and they use 7815 and 7915 regulators and then doing some trickery to get the voltage at the edge connector to be + 16v.  They are using basically the same amperage transformer as I am.  They use 4700uf main power filtering caps before the regulators.  I have been using 2200uf caps so I upped mine to 6800uf without success. 

I then altered my whole circuit to work around some fixed regulators by NTE.... I used nte because that is all that was at the store by me.  They  I set it up so that it was spitting out + 15v.  The module STILL didnt want to behave correctly.  I wonder if i need some proper 7815 and 7915s by texas or STM. 

I havent yet measured the module's current draw by putting my dmm in series with a module break out system yet.  I think I'll try that next. 

This is annoying!

 

[bruce0]

Data please

Measure the rails at or near the module or rack and let us know what it is before during and after the power up cycle.  Use an oscilloscope so we can see deflection during power up if possible.  Sound like something is drawing too much current and sagging and the relays are dropping during the sag... then repeat, but data is what is needed.

The api you say is at +-16v, perhaps you need to be at that same voltage.

 

[Harpo]

Your 7815/7915 vregs might be hanging on startup. Two (maybe 1N400x) diodes between V+/0V and 0V/V-, diodes anode side to the lower voltage rail should fix this. Upping the rail voltage to about +/-16V is easiest done with a standoff diode between 0V and the 78xx/79xx vregs (center) reference pin.
Using the correct value of the Iset resistors at your 317/337 config to draw the vregs min.current draw for correct operation and adapting for the corresponding voltage setting resistor might have fixed your 1st version as well when the power on delay of your vintage design unit might prevent this min.load requirement.

 

[wkbdgeorge]

Hi Bruce and Harpo,

I went back to my 317/337 version.
Voltages without module hooked up are 16.24 and -16.35
When powered on, nothing changes until I press in one of the lighted push switches and thats when the relays start firing and the led switches turn on and off. 

When no switches are engaged, no leds are lit.  That part is normal.

When the module is freaking out, both voltage rails swing towards 0v on my dmm, but I dont see anything on my oscilloscope.  The drop isnt much.  On the positive rail for example, my meter shows 16.18 as the lowest it drops.  The numbers are constantly jumping.

Forgive my ignorance, but I'm not sure I know what you mean Harpo by "Iset" resistors.  Do you mean the lower value resistor that is between the out pin and the adjust pin?  If so, that is 243.  Is that too large..... or to small? The resistor that goes from the adjust pin to ground(0v) is 2.87k. 

Thanks guys.  What do you think?

 

[pucho812]

I think you need to post a schematic of the power supply you are using... start there...

 

[wkbdgeorge]

Ok, ive never posted a schematic to a forum before so hopefully its resolution is sufficient.

Its basically this... disregard all the traces that go to circles.

 

[bruce0]

I know the datasheet puts a 10uf cap on the rails after the regulator (I think it is C13 in your schematic, it is next to C7, but I can't read the number).

Anyway you have those on both rails.  Most 500 series modules put some reservoir caps on the incoming rails, these provide additional isolation of the module from other things happening in the 500 rack, and essentially they are in parallel with those 10uf caps you have there.  Usually they are 47uf to 100uf.

The PSU that silent arts did for GDIY 51x racks has 1000uf in that position.  This acts as a reservoir to handle current surges (such as the ones that you have when turning on ). 

Regulation conceptually suffers somewhat from these caps, because the feedback is from before the cap, but in practice I think they are needed.

Now for a rack of 11 modules 1000uf seems adequate (that psu works)

For a single module 100uf seems adequate (modules live with that in addition to the reservoir cap in the main psu).

I think you may fix the problem by putting a bigger cap in that position.



Re: your post - I meant you to use the Oscope as a Volt meter on the rails to see how far they drop.

DVM's are slow, If you see them drop at all on a DVM you could be getting a serious dip.  Given your 10uf reservoir cap (1/100th of what the GDIY511 rack psu uses) I suspect you have a serious dip, low enough for the relay contact to drop, and starting the cycle you have.

 

[bruce0]

Short version: Bigger reservoir caps after the regulator.  Others may disagree.

Other information: Relays have a pick-up voltage (the voltage they see that will definitely cause the contacts to switch) and they also have a drop out voltage (name?) which is the voltage at which the relay contacts definitely go back to the resting position (for single side stable relays, which yours probably is).

So if your voltage dip in the psu goes BELOW or even NEAR the drop out voltage when your relay turns on, it will turn off the relay.  Then the module will try to turn on again...

Click - click - clickity click

 

[wkbdgeorge]

Hi Bruce,
Thanks for the response.  I'll definitely try a larger cap at that spot and report back right away.

I left something off the schematic I realized.  There is a 220uf cap in parallel with that 10uf right before it leaves the pcb.  Also, I tried putting 220uf and 2200uf caps right at the card edge connector before I posted to the forum with no change.  I havent, however tried upping that 10uf right physically next to the regulator. 

Thank you for explaining the goal of the oscilloscope procedure.  I really dig learning as much as possible in this field.  I'm from a recording engineers background gone techy. I'll also attempt to give some feedback as to what I see when I do that test again.

 

[Harpo]

Forgive my ignorance, but I'm not sure I know what you mean Harpo by "Iset" resistors.  Do you mean the lower value resistor that is between the out pin and the adjust pin?  If so, that is 243.  Is that too large..... or to small? The resistor that goes from the adjust pin to ground(0v) is 2.87k.

From NatSemi datasheet, the LM117 needs a min.load current of up to 5mA. The vregs reference voltage 1.25V / Iset 243R=0.005A assures this min.load requirement. One page further from same datasheet you see the min.load current for the NatSemi LM317 of up to 10mA. Other manufacturers (FI Fairchild LM317) want up to 12mA as a min.load current, so either the 243R has to decrease to 1,25V/0.012A=104R (pick a 100R) and decrease your 2K87 accordingly or use an external resistor between Vout and 0V reference voltage of  (16.25V/(0.012A-0.005A)=2K321 (pick a 2K2) to assure min.load consumption when no other load is connected. Same goes for the negative rail LM337.
The 7824 of your module most likely uses the 32V differential of your +/-16V supply as input to regulate down to its internal 24V requirement that leaves the 0V from your supply floating in the middle of nowhere, so the vregs might misbehave when parts requirements are not met.
Your 0.7A max.current transformer connects to a bridge rectifier with center tap config (K=1.8 ), so the available DC current from your 0.7A AC transformer is 0.7A/1.8= 0.388A, 0.194A DC per rail, enough to meet the VPR compliant requirement of a single module.

 

[bruce0]

Hi Bruce,
Thanks for the response.  I'll definitely try a larger cap at that spot and report back right away.

I left something off the schematic I realized.  There is a 220uf cap in parallel with that 10uf right before it leaves the pcb.  Also, I tried putting 220uf and 2200uf caps right at the card edge connector before I posted to the forum with no change.  I havent, however tried upping that 10uf right physically next to the regulator. 

Thank you for explaining the goal of the oscilloscope procedure.  I really dig learning as much as possible in this field.  I'm from a recording engineers background gone techy. I'll also attempt to give some feedback as to what I see when I do that test again.

Well if there is a 220uF cap there, and you tried larger caps on the card, then I am definitely wrong.

Don't worry about the position of the cap (how close it is to the regulator) that should not matter here.

I suggest set the scope for DC, set it to like 50ms or more and connect it to the two rails (trace on on one rail, trace two on the other and then turn it on and see what is happening to your power. 

 

[bruce0]

hmmm, so I was poking around with the dmm a bit more along with my scope from Brian's suggestion.  The module is an open design without a shroud so I was checking some voltages.  I found some weird stuff happening.  Let me know if this is normal.

The voltages at the input of the card are correct and are hitting the correct pins.  As I was poking around some more, I found that the voltage at its 7824 regulator to be NEGATIVE 16v both before and after the regulator.  Also, I was checking the rest of the audio input and output paths and found that at the output of the module, its putting out +16v on its ground pin. 

Could these modules just be faulty?  But why would they work in the API brand rack?  I checked my shadow hills mono gamas again and they dont exhibit this behavior.  Can lack of current make this voltage weirdness happen?

I missed this post ... wow!

I don't see how it could be putting +16V on a ground pin, on a 500 rack the audio ground pin is either tied to chassis and ground or to psu ground.

I suggest that you test ground continuity on the racks you are using.

Some racks allow ground selection via straps or jumpers.  But when wired up, the XLR ground pin should always be connected to either PSU ground or Chassis ground at the power connector.  And the Chassis ground should always be connected to the Power  ground IN THE PSU. 

It occurs to me that you might have left these disconnected in your psu?  And your rack is wired to expect it? 

 

[Harpo]

TKs module is based on a neve 1081, a single rail 24V design. This has input and output floating/galvanic isolation, so 'ground' inside the module is a relative term and is at different potential from the VPS +/-16V supply without any relation to this usual common rail. XLR-pin1 is only a connection to shield/chassis and a current return path for phantom powered mics and has no relation to the modules reference voltage.