Suggested fuse sizes for MB-6582 (8x SID) load?

[m00dawg]

It lives! Almost :) Apart from a few minor quirks, my MB-6582 is mostly working. I'm in the process of slowly adding in PIC and SIDs and, so far, everything is working. Problem is, I think I either mis-judged my fuse sizes or I have a problem because once I have 3x PIC and 6x (6582) SIDs, I start to blow fuses.

I'm using 1A fuses on the 110V side (on + and Neutral). Seems like that should be plenty. When the fuse blows, there's no odd behavior. My LCD does seem to be flaking out a bit but I don't think that's related (it is flaky well before the fuse blows and is regardless of how much PICs and SIDs I have).

Thoughts? I don't want to start putting bigger fuses in the thing until I know it's safe to do so :) I mean the fuse is there for a reason.

Thanks!

[Wilba]

You need a bigger fuse.

8x 6582A SIDs will draw 800mA on the 5V supply alone.

Add to this an LCD which draws perhaps 250mA and all the chips, LEDs and such... :)

3A seems like a more sensible number... the C64 bricks supply 5V DC 1.5A and 9V AC 1A (at least the good ones I use supply that).

[m00dawg]

Yikes that means I'll need something bigger than a simple 7805 it sounds like, which means I have bigger issues than just a fuse :) I thought the SIDs drew more off the 9V than the 5V? Hmm...

Well thanks for the help as always Wilba! I'll go grab some 3amp fuses and see how that fares.

[Wilba]

I think these C64 PSU bricks just have a 7805 in them, with a huge heatsink, so it can deliver the 1.5A. I'm pretty sure I recall it was a TO-220 package regulator.

If you power the LCD from the 9V supply, this would help reduce current load on a 7805.

In short, just break the 5V to B+ connection and replace with 9V to B+ connection.

See more about it in this post: http://www.midibox.org/forum/index.php/topic,13492.msg118498.html#msg118498

[seppoman]

as the fuses are on the primary (110v) side, 1 A should really be plenty. That's 880 Watts! Even if the efficiency of the transformer is not very good, that's still maybe 300-400 Watts before the regulator, that'd be at least 20A on the 9V and the 5V rail, or more than that on one of them. I think the standard fuse of the C64  brick is 300mA, and that's with plenty of headroom. So I suppose there's something seriously wrong, either the psu is broken or you've got a short somewhere. Are you sure no part on the pcb or the psu itself are getting seriously hot?

S

[Wilba]

Ooof... I missed that bit about primary side  :-[

[seppoman]

uhm... never post before the first morning coffee is finished ;) I don't know how the factor 8 came to my mind, of course 110v/1A is 110 Watts, not 880 :D - still, with that fuse you should be able to draw at least 2.5 A on each rail before it faints out, so better check what's wrong.

S

[m00dawg]

Haha I was wondering about that :) Of course, stupid me, I should've just done the math to figure out the wattage that would have been.

No part of the PSU seems to be getting hot that I can tell, other than the regulators which are attached to a dual TO-220 heatsink. The sink is warm to slightly uncomfortable to the touch, but isn't hot enough to burn skin or anything. I haven't touched the caps or regulators yet though, but, at least last I checked, they weren't in backwards :) There's not a lot of parts on a PSU that could go wrong, though, beyond the regulators?

My guess is that it is in the PSU itself though. Other than the LCD, I don't notice any flakyness with the MB-6852 itself. Fully loaded with all the PICs and SIDs, power isn't supplied to the MB-6582 for more than a few seconds. With 6 SID, 3 PICs, I have just enough time to test audio from the 2 SIDs on the 3rd PIC by playing a few patches and that's about it.

One thing I did notice, however, was that there is a small audible buzz I don't normally hear with only 2 SIDs stuffed. It seems to get louder progressively until the fuse blows, so I have a feeling that's connected somehow.

[cavey]

You sure there isn't a short-circuit somewhere? Stray blob of soldering left somewhere? Dangling bit of wire, flaky soldering joint?

How about testing the PSU separately, hooking it up to a breadboard orso, just to make sure it really isn't the PSU? (or really is the PSU, depending on what you want to confirm)

[m00dawg]

Well, it only happens after adding in the 3rd set of SIDs so it could be something with the MB-6582 itself. Doing a load test is a good idea, although I'll need to figure out how to do that in a way that simulates load form the 6582. Oddly enough, the 6582 is behaving correctly, again, other than the flaky LCD, which is flaky no matter how many SIDs I've got stuff into the 6582.

I'll keep playing around with it and see what I come up with.

[Wilba]

It certainly sounds like it's failing under load... the fact that it happens after the 3rd set basically means when your SID load is over 400mA then it fails. My guess is there's nothing wrong with the PCB itself, or I should say it's pretty unlikely you have a short that's causing the PSU to fail exactly when the SID load is over 400mA.

What exactly are you using to power the MB-6582 PCB?

[m00dawg]

It's my own monster basically. I've got a 9VAC transformer that I take and rectify and supply that as input separately to a 7809 and 7805 that I have heatsinked. Nothing really special, except that I had the luxury of putting fuses on the AC side to avoid killing my xformer (at least that was the plan). I want to make this into a more refined design by adding in resettable fuses (9VAC side) getting a printed board, etc.

Right now the regulation is done on a small protoboard, which does mean the possibility for bad ground or shorts is possible. I'm sort of wondering if grounding is an issue here because of the increasing audible hum I hear (when I have the MB-6582 fully stuffed) over the SIDs shortly before the fuse blows.

I had tied together the earth GND prong from the AC side to the ground of the board and MB-6582. I assume that was the safest way to do it since the PSU case is just plastic. Could that possibly be it? I was going to test unhooking earth GND to see what that did.

I should point out too that, though the heatsink on those regulators isn't ginormous, it's not small either. It does get warm to slightly hot, but not like skin searing hot. So I don't think the regulators are to blame, at least in terms of heat dissipation.

[seppoman]

"GND prong from the AC side" - do you have a 3 contact wall plug and you connected the safety GND to the synth's GND? If yes, you should definitely remove that connection! The GND of a rectified AC supply is just some voltage reference that doesn't need to be on the same potential as the safety GND, so there's probably current flowing via this connection (if you're curious, disconnect and measure AC and DC voltage and current).

As long as the PSU is in a separate box that is made of plastic, there's no sense in connecting earth GND to anything.

S

[m00dawg]

Yes it is a three prong. I'll go ahead and disconnect it and see what that does, although it does kinda bug me I don't have that thing connected anywhere. Isn't it supposed to protect you from ground faults and the like?

[seppoman]

it's supposed to protect the user in the unlikely event of a live AC wire touching a metal case. But if a case is made completely out of plasic, or even the PSU is in a separate case, you will gain no safety at all but only introduce troubles by this connection. Earth should never be connected to the system GND, only to a metal case if there is one. There's billions of all plastic wall wart PSUs in use without such a connection, so I guess the chance of a malfunction that is severe enough to cause a connection between the AC side and the output is really minimal, especially with traditional (transformer) ones. Just make sure that the transformer and the connecting wires are held firmly in place, i.e. use hot glue, cable ties, epoxy, whatever to fix the wires :)

[m00dawg]

it's supposed to protect the user in the unlikely event of a live AC wire touching a metal case. But if a case is made completely out of plasic, or even the PSU is in a separate case, you will gain no safety at all but only introduce troubles by this connection. Earth should never be connected to the system GND, only to a metal case if there is one. There's billions of all plastic wall wart PSUs in use without such a connection, so I guess the chance of a malfunction that is severe enough to cause a connection between the AC side and the output is really minimal, especially with traditional (transformer) ones. Just make sure that the transformer and the connecting wires are held firmly in place, i.e. use hot glue, cable ties, epoxy, whatever to fix the wires :)

Haha noted. For now, though, I'm using electrical tape :) Seems to be working well enough - all the live stuff is insulated by tape or crimp connectors. I just be careful when moving it. I plan on putting it in a better case and will actually glue and screw things down.

[m00dawg]

Well...I feel dumb :) I took my PSU apart this last weekend while I switched out the power socket (added in one with a switch since I haven't installed a switch on my MB-6852 yet). Turns out, I had the fuses hooked into the 9VAC side. Whoops. I put them back over to the 120VAC side and now things work as advertised.

I did find out that my heatsink is woefully undersized :/ The sink was up to 70C so yeah, that's a bit hot :)

My LCD is still flaky though. It's weird - the LCD shows the MIOS boot screen without issue. Once it gets to the patch screen, however, everything goes south. The screen blanks and if I try to change patches it just shows random characters. Sometimes it will sort of reset and I can see things, but most of the time is just garbage. Doesn't seem to be a problem with the LCD since the boot screen always comes up properly. Tried re-uplolading the MB-SID application but to no avail.

Other than these two issues, however, my MB-6852 is ready to rock! All 8 SIDs appear to be working without issue!

[seppoman]

so did you try disconnecting the earth - GND?

[m00dawg]

Yes. I forgot to mention that when I updated the post and didn't remember until I was already heading off to sleepyville. I did indeed disconnect GND. Given the discussion, I have no plans to reconnect it though I still need to cut the wire down just to make sure it doesn't short with anything in the box.

My next plan is to see if I can get my old, larger, heatsink onto my newer power board. That or I will hookup my old board the sink is currently connected to in order to see if that helps. The heatsink is from a motherboard northbridge so I'm thinking if that can't handle it then I am probably looking at a redesign.

I noticed Wilba used some power diodes for sammichSID to step the 9V from the 7809 down a bit before it goes into the 7805. I might try tha as well. Currently, the 7805 is being supplied by the rectified voltage which means it's getting 12V! :P In previous designed I fed the 7805 with the output of the 7809, but that still should throw off quite a bit of heat and I think I'd be close to overloading the 7809 potentially.

[cavey]

Maybe your LCD is flaky because of 4bit/8bit troubles in the LCD driver (either on the software side, or on the hardware side?) Glad to hear you found the solution to your fuse troubles :D

[m00dawg]

Maybe your LCD is flaky because of 4bit/8bit troubles in the LCD driver (either on the software side, or on the hardware side?) Glad to hear you found the solution to your fuse troubles :D

I thought about that, although I sure hope it isn't since it's a CrystalFontz, the same LCD recommended on the MB-6852 base construction guide on the Wiki. But it can't hurt to try I suppose. Could be the LCD might be flaky itself, but it does seem odd that it always shows the MidiBox banner without issues *shrug*

I need to look into that as well as try Wilba's suggestion of running the LCD off the 9V, which should help my power situation as well.

[Wilba]

I noticed Wilba used some power diodes for SammichSID to step the 9V from the 7809 down a bit before it goes into the 7805. I might try tha as well.

That only works for low current situations... each diode is dropping 0.6V-0.7V (I forget exactly) and also dissipating heat. It is a bit of a hack that TK suggested, and it's better than using 1W resistors, which also dissipate heat, but their voltage drop depends on load, whereas the diode is a fairly constant voltage drop. I can get away with it because the current through the 7805 is only 300mA, so the heat dissipated by each diode is small, and I want less load on the heatsink because there's not enough space for a bigger one. In your case, a bigger heatsink is a better solution I think.

[m00dawg]

Damn. Oh well, thanks for the advice Wilba! Do you think there's any use in connected the 7805 to the output of the 7809? Seems like all I am doing there is just moving heat around, but it would avoid supplying 12V to the 7805 directly. I don't think that would avoid needing a bigger heatsink though. I'm going to try my old solution again and see if that heatsink can do a better job.

I suppose the absolute worst case is to use two transformers, but I fear that would not only waste a huge amount of space but also get me into grounding trouble all over again (if you remember my failed attempts at using a center-tapped transformer to supply multiple voltages).

[m00dawg]

That only works for low current situations... each diode is dropping 0.6V-0.7V (I forget exactly) and also dissipating heat. It is a bit of a hack that TK suggested, and it's better than using 1W resistors, which also dissipate heat, but their voltage drop depends on load, whereas the diode is a fairly constant voltage drop. I can get away with it because the current through the 7805 is only 300mA, so the heat dissipated by each diode is small, and I want less load on the heatsink because there's not enough space for a bigger one. In your case, a bigger heatsink is a better solution I think.

I was thinking about this and talking with members from my local hardware group - what about using a power resistor? Like one of those huge ceramic square guys (some which heatsinks on them already) capable of handling multiple watts? Yes the voltage isn't constant, but as long as the drop isn't more than 2V above 5V, the 7805 can still work. I can have even less if I use a low-dropout regulator, which only requires .7-1V or so above the output. I'm still throwing out the excess voltage in heat, but it's not all through the 7805.

Thoughts?

[Wilba]

Check out NorthernLightX's PSU design: http://www.midibox.org/dokuwiki/doku.php?id=northernlightx