SLA battery charger using UC2906

Thread Starter

Zaotron

Joined Mar 8, 2013
4
Hello,

I am having issues with getting the pass transistors to turn on when using the UC2906/UC3906. My input supply is set at 24V and I am attempting to charge a Sealed Lead Acid battery(12V batter with 105 Amp/hours). My shunt resistor is set to .025Ω so that the maximum current (Imax) is 10A. I used the datasheet for the bq24450(as specified in other topics for the UC3906) to select the external pass transistor. I have chosen the external Quasi-Darlington to supply the 10A needed for charging. For Q1, I have selected a PNP BJT transistor(TIP30). For Q2, I have selected a NPN BJT transistor(BUV26).

When I apply power from the supply, I am getting 24V across pins 5 and 6 on the UC2906. , +Vin and GND respectively. So the chip is definitely receiving power, but it is not turning on any of the transistors to provide current to the battery. I am measuring 0V across the shunt resistor, I should be reading .250V when the chip is in the bulk charge state. The battery is at 11.97V, which is well within the bulk charge range.

Could I have selected improper transistor components? Why is the pass element not turning on? The circuit is shown below:




Datasheets can be found here:
UC2906 Datasheet
bq24450 datasheet
UC3906 Application Note
BUV NPN Transistor Datasheet
TIP30 PNP Transistor Datasheet

Thanks!
 
Last edited:

ifixit

Joined Nov 20, 2008
652
  1. It looks to me like Rp should be connected to the battery plus instead of ground.
  2. With 12V across the BUV26 and charge current at 10 Amp, you will fry the device quickly (120W). Keep the input voltage as low as you can, 16V may be enough when charging at 10 Amps. It can be allowed to go higher at lesser currents. It still needs a very good heatsink.
  3. The BUV26 is a switchmode device, select a transistor that is spec'd for linear operation and has a thermal resistance (junc to case) of less than 1°C/W.
Regards,
Ifixit
 

Thread Starter

Zaotron

Joined Mar 8, 2013
4
Thank you very much for the advice! I just had a couple more questions and comments about this circuit.

1. The app note is showing that Rp should be connected to ground due to an internal NPN transistor in the UC2906.

Could the appnote be incorrect?

2. Thanks for the advice on the heatsink, I definitely made sure to attach them to the transistors and the current limiting sense resistor.

3. I'm looking into getting a different transistor to replace the BUV26. Do you think a TIP3055 would do the trick?
I would pick a base resistor of 10 Ohm to supply 10A to the battery.

Also, I was considering using MOSFETs for the pass element as well. Would these following two MOSFETs work for this application?
P-channel MOSFET
N-Channel MOSFET

Would it be wise to put a 100 Ohm resistor on the gate of these MOSFETs?

I think I may be selecting the incorrect transistors for this application, thus, the pass element is not turning on. Any advice on selecting a good pair of PNP and NPN transistors to pass 10A?

Thanks again!
 

BillO

Joined Nov 24, 2008
999
Can you provide a link to that app sheet? You are using the UC2906 in a way I have not seen before. Not saying it's incorrect, but I've just not seen it done like this.
 

ifixit

Joined Nov 20, 2008
652
Hi,
  1. The BQ24450 has a better block diagram and shows a current source where there needs to be one for a circuit to work with Rp connected to ground. The UC2906 appnote does not show that. So in that case leave Rp connected as you have it. This allows you to have a lower drop-out voltage across the external series transistor.
  2. Power dissapation is a big issue. Watts = voltage across the series transistor X current. e.g. 5 Volts X 10 Amps = 50 Watts. The junction temperature rise is; θjc X Power). e.g. 1.4°C/W X 50W = 70°C. Way too high for long life.
  3. The TIP3055 gain at 10A is 5 minimum. You need 2A of base current. The TIP30 is only spec'd for 1A. You shouldn't need base resistors. I'll see if I can find a better choice. You need a gain of 20 min, and low θjc.
Regards,
Ifixit
 

BC107C

Joined Apr 3, 2011
14
If you have access to a reliable distributor look for MJ11033G, a 300W PNP Darlington with an overall gain of minimum 1k at 50A and >8K @ 10A, giving you a max 10mA through the chip.It is about $15.
The schematic poses a big threat to the BE junction of the series pass transistor(s) when you remove the power having the battery connected.Please add some diodes, if not in paralel to CE for the pass regulator at least antiparalel to the BE junctions.
Keep the compensation in the very low freq range , 47n to .22u on 200 to 560ohm series resistor.

At 10 A 12V out, 24V input the transistor will need to disipate a lot, a lot of heat.At this level without a fan on the heatsink you might have some problems.If you still want to try your own darlington do the math before selecting the resistors.For example, on a TIP3055 the max base resistor is 2.7ohm and I would not install any.
for the PNP I would use BDW24.Other combinations are also Ok, including FETs and so on.
 

Thread Starter

Zaotron

Joined Mar 8, 2013
4
Thank you for the advice on the MJ11033G, it seems like it is definitely beefy enough, but I'd like to stay away from using something with a TO-3 package and would prefer using TO-220 packages because they are much easier to mount on my prototype board. Not sure if I'm right though but, I think that the internal transistor that runs from Pin 16 to Pin 15 may limit the current on a single PNP darlington. I believe this is why they specify using a Szilakai Pair(Quasi-Darlington) for anything sourcing more than 1 A.

I am using an adjustable power supply for the input voltage, so I can lower the voltage to about 18V, which should cover the drop from the transistors and lower the overall power consumption. If heat issues arise, I think I will drop the maximum current down to 5A instead.

I would still like to create a pass element using a smaller PNP and a larger NPN.

I was thinking about using the TIP32 for the PNP

and either a BUL743 - this has a 1.25 °C/W thermal resistance and a min. gain of ~5-10 @ 10A

or a D44H8/D44H11 - these don't have as good as thermal resistance 2.5 °C/W, but the gain is 30 at 10 A.

Would these options work?
 
Top