Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Problem with linear power supply, can anyone help?

Status
Not open for further replies.

CarlosFandango

New Member
Dear All, I'm hoping that someone with more power supply design experience can help me out, I'm OK with digital stuff but this one is foxing me a bit. I'd like to apologise in advance for the scruffy hand drawn schematic attached but this was a quick and easy way of doing it...

This is a power supply design which I believe to be quite common, and it's not my own - I found it on the net, it's quite commonly replicated. The PSU supplies a microcontroller plus associated peripheral components, which draws about 3-400mA maximum. It's intended for use on a vehicle which I know is an electrically harsh environment, however the design as shown works fine on most vehicles - it survives all relay switching, wiper motor actuation, fuel pumps etc. without problems. I added the TVS as shown which works well and cured some initial problems.

However I have fitted this to a vehicle with a dynamo based charging system, and it can't tolerate that at all. I've looked at the vehicle's electrical system with a scope, and there are some pretty nasty spikes coming out of that generator; I should say that I believe this to be quite normal for the vehicle in question (a Harley Davidson Ironhead Sportster, actually). These are making their way through to the 5V side quite easily and are causing my MCU to continuously reset.

I've tried a few things so far that have worked for me in the past: I have fitted ferrite beads to the power lines, and I have wound the power lines around a larger ferrite. This does reduce my spiking to some extent but does not alleviate the problem sufficiently to allow the MCU to function reliably.

So what I'm looking for is a method for modifying this circuit, if possible, to provide some kind of protection against this kind of noise. I've found references to pi (LC) filters on this site, and others - but the information provided is so variable, I'm having a hard time working through it. I figured now was the time to ask for some aid!

-CF
 

Attachments

  • PSU schematic.jpg
    PSU schematic.jpg
    64.9 KB · Views: 473
Stick a couple of 1 to 2 ohm resistors and a 100uH inductor in series with the line input and a couple of smaller caps across the main input smoothing cap (100nF, 1uf, 10uF) and see how much that helps.
 
From your userid and the quick reply, would I be right in guessing you've had some experience in this?! OK - sounds good. I'll have to order in a few inductors and the bike in question isn't accessible for a few days, but you've given me something to work with here. I'm very grateful, let's hope it solves my problem!

By the way, I did at one point consider a switching regulator for higher efficiency; the 7805, obviously, gets hot in this application. But whether hot or not fixed linears are very sturdy in my experience, and, after looking into non-linears a bit I started to worry about noise from the switcher - have you any experience of that by any chance? I've never used them myself.

-CF
 
Stick a couple of 1 to 2 ohm resistors and a 100uH inductor in series with the line input and a couple of smaller caps across the main input smoothing cap (100nF, 1uf, 10uF) and see how much that helps.
A HF pie will not help. Vehicle power busses are notoriously full of LF noise. **broken link removed**

Dan
 
You might try a noise filter such as this **broken link removed**.
 
A HF pie will not help. Vehicle power busses are notoriously full of LF noise. **broken link removed**

Dan

Hmm, yes... thanks for the link, some good info there. Well, all I can hope for I guess is that it's at least in part an HF problem. On the scope (which isn't a storage scope unfortunately, or I'd know more) there seem to exist a number of unpleasant artefacts. As I say, this device works fine in alternator-equipped vehicles, it just doesn't tolerate the dynamo - I say 'dynamo' but it is in fact a strange kind of hybrid affair; regardless, it's chucking out all kinds of junk. Might be that I'm simply unable to get it to work, but if I can then I'd be very confident about most potential installations!

-CF
 
You said that the regulator gets hot. If its heatsink is not big enough then the regulator shuts down when it gets too hot.
 
You might try a noise filter such as this **broken link removed**.

Yes, I have seen this kind of thing for car stereos etc. I think these are just low pass LRC filters. I might give it a try, it wouldn't be a solution in the end though as everything has to fit within an enclosure (this won't, and for reasons that would complicate this discussion, I would actually need two...) - but it could point a way foward. Worth trying anyway for sure - thanks for the link...!
 
You said that the regulator gets hot. If its heatsink is not big enough then the regulator shuts down when it gets too hot.

I understand this issue; the reg is on a small heatsink but in a 30 celcius environment test reached 61 celcius at the sink. Shutdown for a 7805 is at 125 celcius, I don't think this is the problem. It's definitely junk getting through to the low tension side of the reg....
 
What does D2 do?

The output voltage will be 5.6V and the regulation will be worse with D2.
 
The output voltage at the positive side of C2 is indeed 5.6/5.7V because D2 is there. The diodes D1 and D3 plus C1 - C3 form a 'diode capacitor filter'. I'm taking my supply from the positive end of C3 - there is a voltage drop across D3 however of 0.7V so to bring the voltage back to 5V proper at C3 positive end, D2 is included. I hope that explains its purpose. D2 is not strictly required as the remainder of the circuit can operate off 6V; but this is the absolute max, and I don't really want to be anywhere near this.
 
D3 will not cancel the effect of D2 because it will have a higher voltage drop due to the higher current passing though it.

I don't see how this helps things, if you want to protect the regulator then it's better to connect a diode from the input to output side (- going to input).
 
I agree with that - I should have really said, "I hope that explains its INTENDED purpose" in so far as it does to an extent. The measured output at 400mA is about 5.2V in fact, whereas it would otherwise be higher. This isn't my design you understand, and clearly it isn't working - hence my current problem!

I found a design for a PSU in the Megasquirt schematic (an enthusiast fuel injection computer by Bowling and Grippo) that has the diode you suggest. That would protect the regulator, but I don't see exactly how it would improve noise rejection? I'm assuming it wouln't in fact do that?
 
Hi CarlosFandango,

you have a lot of unnecessary components in your design and omitted some very important ones. Very important components for use with a voltage regulator are decoupling capacitors (100nF) at the input and output. They should be placed as close as possible to the IC-pins. Their purpose is to prevent regulator oscillations. Regulators might already get hot and switch off due to oscillations without any load connected.

Here is a power supply I designed for use with an MCU to control motorbyce handle heaters.

D1 might be omitted if there is absolutely no risk of false polarity connection. L1 must be made according to load current. 47 to 100µH will be sufficient. D2 and D3 are fast overvoltage protection diodes. They can handle spikes up to 1,500V and the selected breakdown voltage is 15V. The unipolar types are available for 12V breakdown while the bipolar types start at 15V (manufacturer of the diodes: ST-microelectronics).

For C1 you might select any capacity suitable for the expected load current flow. 1,000µF is a good value for 1A. Leave C2 and C3 at a value of 100nF each and connect any electrolytic capacitor at the output to prevent output oscillations caused by load current changes.

Regards

Boncuk
 

Attachments

  • POWER-SUPPLY.jpg
    POWER-SUPPLY.jpg
    23 KB · Views: 175
Thanks for the reply, a couple of things though: this isn't my original design - I owned up to that already! I'm not sure I agree with your comments regarding C2 and C3. Early versions of the 78xx series certainly needed these components but later versions don't require them, at least not according to the datasheet, and indeed no oscillations are observable whether under load or not. Don't get me wrong though, I'm not dismissing this as wrong outright - I intend to try all of the suggestions given so far, as some might obviously work.

Fabrication of inductors is not really possible either, since all components have to be (ideally, anyway) sourceable ex-stock. But there are inductors meeting required value and load characteristics are there not? Why did you make yours? If there's a good reason for that, I'd love to know what it was.

Your D2/D3 combo is similar to the TVS I have fitted already, although the one I'm using is unipolar. I agree a bipolar substitute might work better in retrospect; I'll certainly give that a go too.

What vehicle was your design fitted to? The one I have now has worked on all Japanese and modern H-Ds with alternators, and all cars that it has been fitted to so far, without problems. It's just this Ironhead sporty I'm having trouble with. I'm keen to find a solution to it however, the reasoning behind this being that if I can get a design that works on this vehicle then it will almost certainly survive anything.

The alternative is just to state that the device isn't suitable for fitting to dynamo equipped bikes / cars. I'm close to that 'solution' at the moment...

-CF

PS by the way, there is a massive risk that this thing will be connected incorrectly - the guys fitting them are bikers, not electronic techs., so I will definitely leave D1!! ;-)
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top