What happens when a chipamp loses a rail?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm going to build a basic non-inverting LM3875 amp to go on my electric bike. It'll be run off of a pair of 36V packs run in series. Easy right?

The catch is that they're lithium packs which each have their own battery management board. When a pack's voltage gets too low, it shuts off. Obviously they won't both shut off simultaneously, so one of the rails will disappear.

Each pack is bypassed with a diode to protect it from seeing the whole 72ish volts when it opens. I think this will hold the failed rail to +/-0.6V

how will the amp respond to this?

An externally hosted image should be here but it was not working when we last tested it.
 
Thanks, That's good to know.

I wonder if the diode will make a difference though. It should still hold the rail slightly away from ground, and that might be enough to keep it from flying to the other rail.

Obviously any sound still going to the input would be severely clipped, since the rails would be at +36v / -0.6v or vice versa.
 
When one Rail fails it puts out a lot of DC voltage to the output and stop working , it could potentially burn out your speakers .....

Well you could solve that problem by doing the single supply version that is in the Datasheet , that way you don"t have to worry about frying your speakers when one of the Rails fails ......

just a thought ....
 
When one Rail fails it puts out a lot of DC voltage to the output and stop working , it could potentially burn out your speakers .....
Yeah, obviously this can happen, but I think that might be when the rail actually falls past ground.

Peter's testing has shown that you don't necessarily get huge DC. In his case, losing the positive rail left him with near 0V, losing the negative rail flew the output up to the + rail.

I'm thinking that a simple diode may prevent the rail from actually reaching ground, as long as there's some kind of load present. Anyone think it'll work?
I'll probably give it a try with a dummy load/garbage speaker and see.

Well you could solve that problem by doing the single supply version that is in the Datasheet , that way you don"t have to worry about frying your speakers when one of the Rails fails ......

Yeah. This was my first thought as well, but it does add a lot of extra parts, including some pretty big caps. I'll fall back on that if I can't find a better way, but since this will be mounted to my bicycle, I want to keep it as small as possible.

-Nick
 
You could always add a protection circuit that turns off the amp when DC shows up on its output.

You could also make a similar circuit to shut off both supplies and the speakers, the moment you lose one rail. It could actually be a lot simpler to make.

Just decide which extra parts to go for. There are some DC protection boards on Ebay fot about 10USD though, that's a price that's hard to beat :) I use them all the time.
 
You could use positive rail (first battery) to be voltage reference for the negative rail's (second battery) undervoltage treshold made by resistors as voltage dividers, and vice versa. So the first battery that goes undervoltage will be detected by the other battery's curcuit and the amplifier should be shut down. So you will need one curcuit for each of the two batteries that turns off amplifier if its neighbour battery voltage go below treshold. Good thing is, a curcuit like that will keep on discharging your batteries completely (with time) before recharging them and battery life gets extended. But i would use a switch to disconnect the amplifier from the batteries when its not in use, except when one of the batteries voltage treshold are too low for amplifier to work. Then its better leaved on to discharge the batteries before charging. Maybe two cheap and very low power opamps/comparators with resistors could be used. You could always change inputs (+/-) of opamp to get desired output for both positive and negative rail's opamp/treshold. Another important thing is that the amplifiers shutdown must be made active low (amplifier is shut down when no voltage reference is present at shutdown pin, because if it isnt it will turn the amplifier on again as voltage become low enough.
 
You could use positive rail (first battery) to be voltage reference for the negative rail's (second battery) undervoltage treshold made by resistors as voltage dividers, and vice versa.
Yes, something like this would work, but I'm trying to minimize part count, and I'm not sure it'll be necessary. I'll try it when I have time and see what happens.

Good thing is, a curcuit like that will keep on discharging your batteries completely (with time) before recharging them and battery life gets extended.
This is totally a myth. Running batteries down does not extend their life and, in fact, will often greatly reduce it.
I'm using LiFePO4 packs which shut themselves down for exactly that reason. If you run them down much below 2 volts, they are permanently damaged.

That myth came about from a combination of 3 things.
1. "memory" of NiCd. It is a real effect, but is not really at all like most people understand it. Not really a problem under normal battery use.
2. Electronic battery gauges on some electronics, which need to be run down occasionally to calibrate correctly. That won't increase your battery life either, it'll just make the gauge read more accurately.
3. Crappy batteries are common. Some of them don't last very many cycles (because they're junk), and people assume that it's from the way they were charged/discharged etc.

In general I treat batteries like this:
Lead Acid/AGM/Gel cell, etc: Charge frequently or float with an automatic maintenance charger, and keep discharges as shallow as possible for best life.
NiCd: Charge after using, charge occasionally if not in use. They're much better able to handle fairly deep discharges, but don't run them too flat. There's no need to empty before recharging.
NiMH: Pretty much the same as NiCd.
Lithium (most chemistries): Charge anytime. For longest life, store half charged.
High temperatures reduce the lifespan. No need to empty before recharging. Over discharging cells will permanently damage them. (Some chemistries will stop working, some may cause a fire/explosion hazard on next charge)
 
You could use positive rail (first battery) to be voltage reference for the negative rail's (second battery) undervoltage treshold made by resistors as voltage dividers, and vice versa.
Yes, something like this would work, but I'm trying to minimize part count, and I'm not sure it'll be necessary. I'll try it when I have time and see what happens.

Good thing is, a curcuit like that will keep on discharging your batteries completely (with time) before recharging them and battery life gets extended.
This is totally a myth. Running batteries down does not extend their life and, in fact, will often greatly reduce it.
I'm using LiFePO4 packs which shut themselves down for exactly that reason. If you run them down much below 2 volts, they are permanently damaged.

That myth came about from a combination of 3 things.
1. "memory" of NiCd. It is a real effect, but is not really at all like most people understand it. Not really a problem under normal battery use.
2. Electronic battery gauges on some electronics, which need to be run down occasionally to calibrate correctly. That won't increase your battery life either, it'll just make the gauge read more accurately.
3. Crappy batteries are common. Some of them don't last very many cycles (because they're junk), and people assume that it's from the way they were charged/discharged etc.

In general I treat batteries like this:
Lead Acid/AGM/Gel cell, etc: Charge frequently or float with an automatic maintenance charger, and keep discharges as shallow as possible for best life.
NiCd: Charge after using, charge occasionally if not in use. They're much better able to handle fairly deep discharges, but don't run them too flat. There's no need to empty before recharging.
NiMH: Pretty much the same as NiCd.
Lithium (most chemistries): Charge anytime. For longest life, store half charged.
High temperatures reduce the lifespan. No need to empty before recharging. Over discharging cells will permanently damage them. (Some chemistries will stop working, some may cause a fire/explosion hazard on next charge)
 
I independently confirmed Peter's measurements. I got the same results. Lose the + rail, you get about -1.5VDC on the output, no biggie. Lose the - rail, and you get +20VDC.

So, I threw together a circuit in an online simulator that seems to work (on the computer). Anyone see any problems with it, before I order some fets and build it IRL? (Besides the P-channel fet being backwards. I noticed that.)

Any thoughts?

protcircuit.png


PS: The 100ohm resistors on the right are the load.
 
It seems to me that you're making this more complicated than it needs to be. A single inexpensive relay, connected across the power rails, with its contacts where you show yours, would do it. If either side fails, the relay de-energizes, killing power to both sides. You may need to use a dropping resistor in series with the relay to get the coil voltage right. You can get some small relays that draw as little as a few milliamps.
 
It seems to me that you're making this more complicated than it needs to be. A single inexpensive relay, connected across the power rails, with its contacts where you show yours, would do it. If either side fails, the relay de-energizes, killing power to both sides. You may need to use a dropping resistor in series with the relay to get the coil voltage right.
look at the data for any relay.
The pull in voltage is usually 60% to 90% of rated voltage
The drop out voltage is usually 10% to 30% of rated voltage.

If you ensure the pull in voltage is reached at mains minimum tolerance, you will find that on nominal mains voltage and when half the relay supply voltage is lost, the relay will not drop out.
 
If you ensure the pull in voltage is reached at mains minimum tolerance, you will find that on nominal mains voltage and when half the relay supply voltage is lost, the relay will not drop out.

Yeah. I figured that one instinctively, which is why I wanted to go with the fets. Also, it always seemed to me that relays are likely to _cause_ a rail failure at some point in their lives.

I simplified the circuit a bit more by just removing the connections directly to the rails and connecting the gates with a single 33k. (It works the same, but will be simpler to wire)

Re: Protection circuits. I thought of that, but I'm trying to keep this as simple as possible. Seeing as it's going to be biamped, I would need 4 DC protection circuits.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.