How do I add a diode to stop shorting out my DC Converter?

[ryan]

I've shorted out two of Lyen's DC-DC Converters so far and I'm eager not to do it again.

Lyen recommends a diode to solve my issues, but I'm an EE novice, so I need help with which diode to buy and where to put it. Ideally I'd like it to be on one of the leads up to the dc-dc converter rather than in the main system. Can you help?

Here's my setup:


The problem came when I attached the opposite battery first without turning the switch off. Instant voltage with the wrong polarity. When I connected them in the right order everything went perfectly. Until I detached and reattached, well, imperfectly.

What do you recommend?

 

[amberwolf]

If you're not drawing much current, you could pull a 1N4007 diode out of an old computer power supply, and just put it in series with either the positive or negative lead to the DC-DC input. If you need lots of current just use several in parallel, with air space around them.

That diode is really common in stuff, so you certainly have some already laying around, and it'll handle up to 1A constant at 1000V. Peaks of 20-30A IIRC. The 4006 down thru 4004 are also common (maybe even moreso) and are 800V down thru 400V.

 

[ryan]

Thanks Amberwolf.

Can you help me understand how a 1A diode is able to handle a pack with 20Ah? I know it has to be in the difference between Ah and A, but I'm fuzzy on how that all works.

Thankfully Lyen threw in a couple diodes to help me out. So while I won't have to head to Radio Shack to figure this out, I still want to understand how it all works.

Also, I assume that I need to put the diodes on the leads heading directly to the converter. Is that right?

And would you recommend soldering all of this to a breadboard? Or just soldering and heat shrinking directly?

 

[neptronix]

There's the missing link.
In the for sale thread, he says it takes only 60v input. You're running 72v, which is probably hot off the charger at ~80v.

 

[ryan]

That's exactly right, nep. So how do you recommend running this converter only off of one of the packs?

Like this?

 

[neptronix]

Well.. you could run it on the first pack of the serial group. That pack is 36v before having the other added to it. You can bust out the multimeter to find out what the first pack is if you're unsure.

The downside to this is that one pack will get a little disbalanced over time. this would make bulk charging the 72v all at once an issue.

another solution is to buy a big 2s pack from hobbyking, use that, and skip the dc-dc convertor entirely. The magicshine will be happy at 8.4v - 6.0v.

 

[amberwolf]

Neptronix is right about the pack voltage (I didn't pay attention to that part, sorry). You could put a bunch of diodes in series, at about 0.7V drop on each one it'd take around 29 of them to "absorb" the difference between hot-off-the-charger 80V and the limit of 60V on the DC-DC input. But running it off only one pack at a time would be a better idea.

To eliminate the imbalance problem, you could switch which pack it's running off of on each ride, either manually or with a toggle switch.

Can you help me understand how a 1A diode is able to handle a pack with 20Ah? I know it has to be in the difference between Ah and A, but I'm fuzzy on how that all works.

It doesn't matter what the capacity of the pack itself is (the 20Ah), as that is just how much power it holds inside. What matters is how much current is actually drawn by the DC-DC converter and parts after it. That's probably less than 1A, but if you're not sure you can always parallel two or three diodes, depending on which ones you use.

Also, I assume that I need to put the diodes on the leads heading directly to the converter. Is that right?

Yes. In series with either the + or - lead (or both, but only one is needed). The band on the diode goes toward the negative, meaning toward the DC-DC if used in the + lead, or toward the battery wire if used in the - lead.

Or just soldering and heat shrinking directly?

I'd just lap-solder it as shown in SpinningMagnets' thread:
http://endless-sphere.com/forums/viewtopic.php?f=16&t=23554

 

[fechter]

Another approach would be to wire the dc-dc directly to the battery terminals so there's no way it can get reverse current. You could have a separate connector straight from the battery for the dc-dc. I hope you have a switch in there somewhere too.

 

[ryan]

Thanks nep, I charge the packs separately as is. So I don't think that will be an issue, and my commute is less than 2 miles in the dark so it's not as if I'm running the batteries dry with this setup. I just want to be safe for the 5 minutes I'm on the bike each night.

I do plan on running off the first pack. When fully charged it's up to 39-40v off the multimeter. I hope that a couple diodes will make it all work for me. I'm also installing a switch here as well, so I can turn the entire converter off while I make all of the connections. Then I can even test what is coming up through the leads before I switch it on and KNOW that I'm dealing with 36-40v with the correct polarity.

Well.. you could run it on the first pack of the serial group. That pack is 36v before having the other added to it. You can bust out the multimeter to find out what the first pack is if you're unsure.

The downside to this is that one pack will get a little disbalanced over time. this would make bulk charging the 72v all at once an issue.

another solution is to buy a big 2s pack from hobbyking, use that, and skip the dc-dc convertor entirely. The magicshine will be happy at 8.4v - 6.0v.

 

[ryan]

Isn't that essentially what I'm doing? Just with a few inches of extra lead? What is the difference between the first diagram above and what you're suggesting?

Another approach would be to wire the dc-dc directly to the battery terminals so there's no way it can get reverse current. You could have a separate connector straight from the battery for the dc-dc. I hope you have a switch in there somewhere too.

 

[fechter]

Ah, all the difference in the world.

Think about what causes the dc-dc to blow up. Looking at your excellent diagram, if the right hand battery is connected first, current will flow backwards through the dc-dc to charge the capacitors in the controller. This capacitor charging current can be very high for a short time and you will frequently see a spark when that connection is made unless a precharge circuit is used. This is definitely enough to blow the dc-dc. If you alway connected the left side battery first, you wouldn't have a problem.

If the dc-dc is tied in directly to the left side battery, connecting the right side one first won't result in reverse current through the dc-dc. This way you could connect things in any order without problems.

Another approach would be to use a large diode across the input to the dc-dc to protect against reverse current. During normal operation, the diode wouldn't do anything, so no losses, but if the connections were made in the wrong order, any reverse current would be carried by the diode, protecting the dc-dc.

 

[ryan]

Thanks fetcher. Here's my plan with my 3rd converter (delivered yesterday)...

Put the converter on a perf board.
Put the diodes (in both directions, as above) on the perf board
Solder the diodes directly to the dc-in pins on the converter.
Wire a rocker switch to the positive wire (from the "left" battery in my diagram)
Install the rocker switch on the project box.
Wire the perf board to the dc in and out wires.
Turn the converter switch off.
Turn the series switch off.
Put the converter into the circuit.
Connect the "left" battery first, positive first.
Connect the "right" pack next.
Turn on converter switch
Verify potentiometer output voltage is 8.4v with multimeter
Turn on series switch
Verify converter output is 8.4v (or curse loudly if I blew another converter)
Plug in lights
Ride like the wind

What do you think?

 

[fechter]

That should work, but really only one diode is necessry in that configuration.

Yet another approach would be much like the way I have my scooter wired. I have the dc-dc tied directly to the motor controller inputs. I have a main relay that kills power to the controller, so kills the dc-dc at the same time.

If you connected the dc-dc to the controller input wires (maybe with a switch), there would not be any way your previous problem would occur. No diode necessary.

If you use a diode or reconfigure the wiring, it should not matter what order you make the connections.

 

[ryan]

Did I just install the diode backwards?

View attachment Screen shot 2011-02-11 at 4.56.29 PM.png

Multimeter on:
A: 39v
B: 39v
C: 0v

It's a little hard to see, but the diode has a gray stripe on the right for the top one and the left for the bottom one.

Note: A connects to B underneath the board. And C connects to V-in.

 

[ryan]

Yes. Clearly. Dumb.

The good news is that I know the diodes work!

 

[ryan]

Reversed the diodes (or rather bought new ones and used those in the reversed position) and everything works! I put it in a project box and everything looks solid, pretty, and it held up to a nice long ride today with day-time lights!

Thanks everyone for your help!