Pic datasheet: WARNING!

[SounDuke]

Thanks to this great community I have learned a lot about PSU's. And I learnt also that datasheets are not always very reliable: some of my custom supplies deliver to core 5.25 (measured with a precise lab tester). This should be no problem because the pics datasheet tells that the pic cannot be harmed under 5.4V. Yesterday one of my pics got fried with 5.25V (desoldered 7805 and jumper on input and output holes), I'm sure that my psu was delivering stable 5.25V. This has made me paranoic, as I'm planning to use some switching PSU's that deliver 5.17V. Can anybody point me in the right direction to step down the 0.17V? I tried to put a load but the voltage is always the same. I became a maniac of stable voltage!

Thanks

[Lall]

Hi,

You could use a LM317 for example, they allow to adapt the output voltage with the use of a simple potentiometer.

That way you can generate a real 5V without any problem. Moreover, you won't have any problem to find some that deliver as much current as the usual 7805.

Best regards,

Lall

[SounDuke]

The problem is that I need at least 2.5A, I fear the lm317 will get too hot. So I was thinking if it is possible to reduce the voltage using diode drop

[Lall]

Hi,

Indeed with 2.5A, you'll need to put two LM317 or use a schematic that was already presented on this forum that has a regulator to provide regulated voltage and another transistor to provide the rest of the current.

Maybe some zener diode could do the job but I don't know how precise they are. I mean that if a 5V zener is precise to 5% then it's potential error will be quite close to what you try to eliminate...

Best regards,

Lall

[TheAncientOne]

The problem is that I need at least 2.5A, I fear the lm317 will get too hot. So I was thinking if it is possible to reduce the voltage using diode drop

Have you though of spliiting the work over 2 or more regulators? If part of the current is for LED's etc, then they aren't that critical, and you can use the expensive precision regulators for your cores. I did an industrial control panel where about 40 LED's were fed from their own variable voltage regulator, so they could be adjusted in brightness to  cope with summer days and winter nights.

A decently big silicon diode will drop about 0.7V, rising with current. for example, a 1N5400 series diode will drop 1 volt  at it's rated 3Amps. There is a problem here though: at 3 Amps 1 Volt drop, the diode will be getting rid of 3 Watts of heat. T_jmax for one of these is 170^oC, and case to air thermal resistance is 53^oC/W, meaning that at 3 Amps and an ambient temp of 20^oC, the junction will start to  overheat. (53x3 + 20 > 170), So I'd use a much higher rated diode. 

A 1N5821 Schottky Diode will drop about 0.5V at 3A, and will only have to lose 1.5W. Doing the same sums, with T_jmax125^oC, and thermal resistance 28^oC/W, gives a reasonable margin, (28x1.5 + 20 = 72 << 125),  though you could split the load over 2 diodes for safety.

(I showed the 'wrong' case to illustrate the pitfalls).

Hope this helps

Mike

[sneakthief]

You could also get one of these 5.0v high-precision voltage regulators from Microchip:

http://www.microchip.com/stellent/idcplg?IdcService=SS_GET_PAGE&nodeId=1335&dDocName=en010617

cheers,

michel

[Flying Panther]

The LM317 can handle 1.5A max, if I'm right. You might want to use a LM350 (3A) or even LT1038 (10A). Don't know how hot they will get though with a 2.5A load. (But since they only need to dissipate 0.25V I think that should be allright).

[TheAncientOne]

I might need to coin a new mnemonic here:  "RTFDS" = "Read The Fantastic Data Sheet"

Only joking - you've been bitten by a well known 'Gotcha' here.

Most regulators have a 'drop out' voltage - in other word the minimum needed difference between input and output voltage, less than which they won't work.

In the case of the LM350 this is 1.5 Volts, dropping to 1.0 volts at the devices maximum working temperature. The only way to use one would be to jack up the initial regulated voltage to (say) 7Volts and spill off the last 2 Volts in the LM350. You'd still need to get rid of 5 watts though - which means some kind of heatsink.

Tweaking small changes in output voltage is a bit of a sneaky one, sometimes.

Mike

[stryd_one]

"RTFDS"

LMAO

[lylehaze]

SounDuke,

So Far, my own experience with Pics has been that they are pretty tough devices. I'm curious about what killed yours.

2.5 Amps is pretty high. Are you by chance switching an inductive load without proper protection? That might create a problem like you have described. If you can tell us a bit more about your project, we might be able to help solve the problem better.

LyleHaze

I checked the 18F452 datasheet:

Operating voltage (should work properly) max 5.5 volts

Absolute maximum (permanent damage at) 7.5 volts

I'm willing to bet that you are exceeding something on

the absolute maximum page of the datasheet. Possibly

voltage, possibly current. Maybe even clamping current.

More details will help. Where is all that power going?

[Flying Panther]

I might need to coin a new mnemonic here:  "RTFDS" = "Read The Fantastic Data Sheet"

Hmmm... Should have thought about the minimum input/output difference.. thnx for correcting that :)

[MRE]

Id have to agree with Haze here..

The PIC has been around a LONG time.. hell, most microcontrollers have. In that time, Im willing to bet that a million micro projects have been feed off the humble 7805 with no over voltage ill effects.

Something else might be wrong here.

A defective regulator?

A defective PIC?  (I have had ONE.)

Or more likely, a design or wiring issue. The number one, two and three causes of dead PICs are:

1: not adhering to source/sink current rules

2: not ADHERING to source/sink current rules

and..  drum roll please

3: not adhering to source/sink current rules.

Now, that is not to say that a design is bad. The design itself could fit the rules precisely, but a misguided screwdriver while power is applied brings the whole thing down.

[SounDuke]

Sorry for renringing the topic, I was very busy so I couldn't visit the forum for a long time.

After reading lylehaze's post, I started to think that I was a little bit fast on saying that the datasheets are wrong (good lesson). I checked my custom psu, and I think that the problem was a defective 7805 (two pics were damaged). Anyway, I'll try to put a lamp and a fuse on the 5,15V output of the laptop switching psu, and see if voltage decreases with lamp+core and lcd load. I'll keep you informed, thank you very much for supporting me

[Lall]

Hi,

I did burn a PIC and a LCD controller most probably as well with a defective PSU. It was a home-made one, never put into a box and the regulator moved enough to get off the PCB tracks and provide the input voltage at its output...

We're currently redesigning our SEQs with Xanboroon and I'm adding some protections like fuses and zener and protection against reverse voltage to make sure that such a thing will not happen again. Actually, it's even more severe in the case of a SID or a FM where the chips are quite difficult to get. I think it's pretty good practice to try to add such protection, it's so easy to make mistakes...

Best regards,

Lall

[MRE]

good call there..

and.. you are totally right about the sid box..

break it, burn it, toast it and finally fix it.. now.. when its just a PIC hitting the bin, and not precious SIDs.

[lylehaze]

@SounDuke

Most PC type power supplies need a minimum load to regulate properly. If it's an off-the-shelf industrial unit you can find out by reading the datasheet, but if you pulled it from a PC you have to figure it out yourself.

I know that some PC parts vendors will sell a dummy load for just that purpose. It used to be the +5 supply that had to be loaded, but I haven't played with that stuff for years. It may have changed with more PC chips running lower voltages now.

Good Luck!

LyleHaze

[SounDuke]

this is definitely scary! I'm a little bit confused about fuses. Thorsten recommends slow blow fuses for mblc, is it the same for mbsid and mb64? ). Slow blow fuses can tolerate a transient overcurrent condition (such as the high starting current of an electric motor), but will open if the overcurrent condition is sustained. At first thought, it sounds like fast fuses should protect better the sids and pics, so could anyone explain me which is the better choice between fast and slow fuses? Thanks ;)

[SounDuke]

all right, I put a 400mA lamp as a load for the 5v 2,5A output of the psu. It stabilized the voltage to perfect 5V, but as I hooked it to core+sid (no lcd, only core and sid), it worked for a little bit then the voltage dropped to 4.96, and the pic had no sufficient voltage to drive the sid. So I removed the lamp and fed the 5,17 V 2,5A to the core, and my MBsid worked ok for an hour of continuous play (at the first try!). I assume that the load with the lamp was about 500mA, so I fear that when I'll hook the lcd and the 3 remaining cores and sid the voltage will fall below 5V.

What do you think? this psu was very cheap, but it's possible that I can't use it. In this case it's really a pity, because although it's a switching psu, the noise level is incredibly low.

???

[lylehaze]

OK, let's try this again.

The PIC should operate properly as long as the voltage

is anywhere between 4.2 and 5.5 volts. The PIC should not

be damaged unless the voltage goes below -0.3 volts, or above +7.5 volts with respect to Vss.

The PIC did NOT fail because the voltage dropped to 4.96 volts. It will run fine at that voltage all day long.

The PIC, like most electronic devices, IS very sensitive to voltage spikes. switching higher loads without proper decoupling can damage a PIC. Switching OFF an inductive load without having the proper parts to absorb the back EMF will definetly kill a PIC. Neither of these "events" can be easily seen on a meter.

Your problem may be your power supply. Your problem may be whatever you are driving with your circuit. If you can show me your circuit, I would be happy to look it over and offer whatever input that I can.

LyleHaze

[SounDuke]

Hi lylehaze, thank you for your support.

he PIC should operate properly as long as the voltage

is anywhere between 4.2 and 5.5 volts. The PIC should not

be damaged unless the voltage goes below -0.3 volts, or above +7.5 volts with respect to Vss.

The PIC did NOT fail because the voltage dropped to 4.96 volts. It will run fine at that voltage all day

long.

I supposed the problem was the low voltage level because when the voltage fell at 4.6V I couldn't receive the starting sysex string from the pic, but maybe the problem is different.

The PIC, like most electronic devices, IS very sensitive to voltage spikes. switching higher loads without proper decoupling can damage a PIC. Switching OFF an inductive load without having the proper parts to absorb the back EMF will definetly kill a PIC. Neither of these "events" can be easily seen on a meter.

I thought that voltage spikes are prevented by the capacitor filters on the core and the 100nF caps close to the power pins of the IC's.  But you're right, let me explain the circuit:

5V 2,5A output of laptop switching PSU connected to a switch, 1A slowblow Fuse and then to core ( to output pins of desoldered rectifier, put the bridge on the holes where usually sit the 7805 (which I have obviously desoldered) input and outputs pins and left the capacitor smoothing section unaltered. Added 100nF capacitors on all of the IC's power pins (except optocoupler).

Since I own a sid module pcb rev2, I carefully followed the information on the special webpage on ucapps with infos on rev 2 pcb version. so the 5V part of the Sid module is supplied through J2 of the core like described in the rev2 --> core  interconnection schematic by TK. the 9V for the 8580 is supplied by the 12v1A output of the laptop switching PSU, I connected only the +12V rail (to switch, fuse and J1 of sid module with rectifier and 7809 stuffed) as the ground comes from J2.

I think that the high load problem (around 500mA the pic doesn't start correctly) arised because of the laptop psu design, which I ignore. Furthermore I don't want to damage the circuit, so if someone notice an error in the circuit I tried to explain, especially regarding protection and adding additional components before the modules,  I'll be glad to hear wise words.

Thank you, and sorry, my learning curve is slow :P

EDIT: as I said, the problem occurred only with the 400mA 6V lamp on the 5V rail before the switch and fuse.

[lylehaze]

OK, that helps.

From the start I have wondered why you need a 2.5 Amp power supply. I was afraid that you were switching relay coils or solenoids or some other inductive loads.

If you have nothing more than what you described on your power supply, we can rule out those problems.

Using a lamp as a power supply load will work, but you need to know that regular lamps draw a LOT more current than usual when they are first turned on. There is a "inrush" current that is needed to get the filament up to temperature.

We still may have a power supply problem. the easiest way to find out would be to put the MB power supply back to "normal" and see how well it works with its own regulator supplying the power. Of course that depends on whether you have a transformer available.

I'm suggesting that we start by getting a "known good" core module first, then if problems appear when you change over to your power supply, you'll have a good idea of why.

Good Luck,

LyleHaze

[SounDuke]

Hi lylehaze, I'm slowly upgrading to step C and other nice personal add on, such as a lcd graphic display and a oled 20x2 display, so I need a lot of power. In addition to this, I wanted a single psu since I can't use c64 psu's (I don't want to use stepper down adaptator when using the mbsid outside europe with 115V input). I used the core in the past with normal 1A power supply. so the core is not a problem. I'll continue building the CS, then I'll see what's the behaviour of the switching PSU with this increased load. I want only to be sure that I don't harm the sid. 

[lylehaze]

@ SounDuke,

I'm sure Sid will appreciate that.  ;)

LyleHaze