[gracky]

After considerations about power supply for my (supposedly) portable amp, which is comprised of 2 AD823s, Ohman's Xfeed, and Jung's discreet buffer, I've tried to build the circuit of PPA battery board by Tangent. So I'd like to thank him first.

It seems to work, at least it's possible to charge batteries. The problems are 1) when it goes to the fast charge mode, some noise is heard via cans, to be specific at the moment when led was gone, so I can't use amp while charging, 2) when the charging is done, if amp's power is off, charger goes to trickle charge mode (led on constantly), however as soon as put the power on, the charger goes back to fast charge mode, so I can't use the amp with wallpower connected. I doubt this is normal operation of the circuit.

Any comment or suggestion will be greatly helpful. Thank you in advance.

My configurations are
8 1.2v AAA 700mha batteries,
R1 240 ohm
R2 about 2.6 ohm
R3 120 ohm
R4/R5 6.8k / 1.5k ohm.
Wallpower 16v/1A (not regulated)

 

[ppl]

to answer your second question first the amp when first powered up appear as a short circuit untill the power supply caps are charged. this triggers the chip into the fast charge mode becaues of the temp drop in wall wart voltage. You May need to get a larger current or higher voltage. wall wart. the noise you talked about i at this time can't comment on as i have not fully evauated the battery board, however i will soon. Hopfully this noise is not normal. I must point out that the battery board was the last part of the PPA that Tangent tested and he did not report any noise problems However i belive this was using only the PPA and the PPA is more tolerent of bad power than most Amps. please post any futher results you obtained stating what other equipment (Amp) you are using

 

[tangent]

Quote:

when it goes to the fast charge mode, some noise is heard via cans

Yes, this was talked about in the thread on the battery board. If you look at the PPA amp schematic, you'll notice that the battery board runs in semi-parallel to the amp board: the amp only drains power from the batteries when the wall power goes away. What you're trying to do is run the battery board directly between the wall supply and the amp -- in series. What you're hearing is the glitch on the rails as the high charging current is momentarily interrupted. You can try to filter that glitch out, but I think it's better to just give up on series operation.

There's an exception to this: if you want to use a simple trickle charger, you can run the pack and charger in series with the amp without problems because you don't get the big current swings like in fast charging mode.

Quote:

as soon as put the power on, the charger goes back to fast charge mode

That's normal. ppl's explanation is probably correct...I have no better theory.

It's only a temporary condition. The chip will fast-charge for the minimum on-time (read the datasheet) and then start checking the batteries to see if they're charged. If they are, it'll go back to trickle charging. This mode of operation is fine as long as you don't toggle the power on and off frequently. If you do that, the minimum fast-charge times will start to overheat the batteries unless you have a temperature sensor hooked up to R6-R8.

 

[Chipko]

EDIT:

I've revised the circuit to take full advantage of the MC33340, just need to make some minor changes and I will have it up.

/EDIT

Still viable from the first post:


The buzz coming from the charger when fast charging can be a result of the voltage to frequency circuit inside the charging circuit, not sure how to solve that.

As for the switching to fast charge when powering up the amp: One solution is to put a good sized capacitor on Vcc to supply it while the amp is filling up it's capacitors the capacitor on the Vsen should also be beefed up if you do this, if Vsen goes out of range it will start to fast charge when it comes back in range. The "correct" option would be to build a slow start circuit for the amp There's a schematic in lm317 datasheet.

 

[Chipko]

Revised the circuit. Couldn't do it with just discrete components so I've thrown in 2 relays from Omron (digikey #Z774-ND) USD2.33 each. They have a 5ms responsetime, well within what is needed and a low power consumption 150mW

http://rocky.digikey.com/WebLib/Omro...Data/G5V-1.pdf

I've based it on Tangents circuit but changed it so that both fast charge and trickle charge is done through the regulator. I've omitted the railsplitters, the power consumption of the chip is low and the transistors and LED consumes very little power, a voltagedivider is easier and won't consume very much power, 30mA or so should be enough, 1mA for each transistor and 2 for a low power LED, it's blinking when fastcharging so anything else is probably annoying.

Vcc needs to be between 3 and 18V, aim for 15ish or so.

All values depends on the batteries you want so choose values on the resistors according to their datasheet. Remember that RF will see a lot of current.

Revised charger PDF

 

[tangent]

Quote:

The buzz coming from the charger when fast charging can be a result of the voltage to frequency circuit inside the charging circuit, not sure how to solve that.

I don't think he's talking about a buzz. I believe he's talking about a periodic click that happens every 1.38 seconds when the charge controller tests the battery's voltage. This is due to a glitch on the power rails as a result of the large current shift as the regulator turns on and off. This is readily audible in most amps when you put the battery circuit in series between the wall supply and the amp.

In the PPA, we don't use that arrangement, for this reason and others. Instead, the wall power makes a "Y" going to the battery board and to one leg of a diode OR bridge. The battery circuit's output ("B+") comes back up to the amp through the other leg of the diode OR bridge. This arrangement stops the glitch from getting to the amp rails. It also solves the problem of how much current you need for trickle charging. Think it through, and you'll see why this is a problem in the series configuration.

Quote:

One solution is to put a good sized capacitor on Vcc to supply it while the amp is filling up it's capacitors the capacitor on the Vsen should also be beefed up if you do this,

That might work. Something to try for v2.

Quote:

The "correct" option would be to build a slow start circuit for the amp There's a schematic in lm317 datasheet.

Do you mean the "Slow Turn-On 15V Regulator"? Yuck. Either you'd have to set the regulator's output voltage to be 2V below the battery pack's minimum voltage and dissipate large amounts of power, or else the regulator would fall out of regulation at some point, adding a bunch of noise to the circuit and providing no benefit.

Another problem is that a slow ramp-up voltage will annoy several good op-amps. That's one reason we ended up rejecting the simple cap multiplier wall power supply idea. It just turned on too slowly to be useful.

Quote:

I've thrown in 2 relays from Omron (digikey #Z774-ND) USD2.33 each.

You've spent $5 on relays and transistors just so you can avoid 7 cents for a diode and resistor in the trickle charge path. What have I missed?

Also, do you not want to use a PNP for Q3? Virtually all of the current will go down through pin 2 or 3 in the charge controller, very little will go into the base of Q3. You want to pull current from the base instead, to avoid this problem.

Quote:

I've omitted the railsplitters

When pin 2 or 3 opens a path to ground, that puts the LED circuit in parallel with Rdiv2, defeating your divider unless you use resistor values about 10x lower than your LED current limiting resistor. That's about 220 ohms each if you use a 4.7K LED resistor like I do. That puts 68mA through the divider, dissipating about 2W!

I thought about what you proposed already, and rejected it for this reason. Yes, a TLE2426 is expensive, but it's far preferrable to the alternative!

 

[Chipko]

Quote:

You've spent $5 on relays and transistors just so you can avoid 7 cents for a diode and resistor in the trickle charge path. What have I missed?

That the tricklecharge is applied all the time in your circuit, that screws with the sensing. It will take longer for the circuit to sense a full charge which will effect the lifespan of the batteries.

The second relay can probably be omitted connecting Q3 directly to the line after the relay1. If Q3 can handle Ic which would equal the fast charge current then it should sink the voltage left of D so that Vsen can do its job, especially if I get around to connect it on the right side of D

Quote:

Also, do you not want to use a PNP for Q3? Virtually all of the current will go down through pin 2 or 3 in the charge controller, very little will go into the base of Q3. You want to pull current from the base instead, to avoid this problem.

The transistor opens when Gate goes high, other times the base is connected to ground keeping the circuit closed.


Quote:

When pin 2 or 3 opens a path to ground, that puts the LED circuit in parallel with Rdiv2, defeating your divider unless you use resistor values about 10x lower than your LED current limiting resistor. That's about 220 ohms each if you use a 4.7K LED resistor like I do. That puts 68mA through the divider, dissipating about 2W!

Good point, another transistor or reusing Q1 so that the base goes to Vcc in series with a resistor. LED in series with RLED between V+ and ground.

With high hfe transistors the resistor values can be kept high. In a configuration like this the currents at the voltage divider should be very low and not vary much why the x10 rule can be atleast halved, probably more after some trial and error.

If you stick with the TLE2426 I suggest a Zener between it and ground. It should lessen the effects of powershortage as the MC33340 will work with a higher Vcc. A voltagedrop over the TLE2426 will be smaller to the MC33340. DeltaVcc=(Vzener-DeltaV+)/2 instead of DeltaVcc=DeltaV+/2. If you calculate the relative drop the differences are even bigger since you have a smaller drop on a higher voltage. Might be a prettier solution than messing with the size of the caps, especially C2.

 

[tangent]

Quote:

the tricklecharge is applied all the time in your circuit, that screws with the sensing

Trickle charge can be uncommonly low in this circuit. It's more of a maintenance current, and also to revive badly depleted batteries before starting fast-charging. I wouldn't recommend C/10 as you would for a plain trickle charger. Instead, 10-20mA for 750mAh cells is plenty here. Given that, is your objection valid? I know I haven't noticed the cells getting really hot at the end of the charge cycle. They barely get warm in my experience.

Quote:

It will take longer for the circuit to sense a full charge which will effect the lifespan of the batteries.

If ultimate battery lifespan is your goal, you shouldn't be fast-charging in the first place. Use a C/10 trickle charger and be sure to turn it off after 12 hours or so. That's the path to better cell lifetime.

Quote:

The transistor opens when Gate goes high

Ah, I should have seen that. I was looking at it backwards.

Quote:

another transistor or reusing Q1 so that the base goes to Vcc in series with a resistor.

Want to draw this for me? I'm not sure what you mean. Also, be sure you account for the fact that Vce will vary based on current through the transistor.

I'm afraid you're going to nearly reinvent the TLE2426 in discretes. What would be the point in that?

Quote:

If you stick with the TLE2426 I suggest a Zener between it and ground. It should lessen the effects of powershortage as the MC33340 will work with a higher Vcc.

The TLE2426es are only needed when the supply voltage is over 20V. Surely 10V is enough to power this chip reliably? And, that's a minimum. More often you'll see 24V being used, just because it's an easy value to find.

 

[Chipko]

Quote:

Trickle charge can be uncommonly low in this circuit. It's more of a maintenance current, and also to revive badly depleted batteries before starting fast-charging. I wouldn't recommend C/10 as you would for a plain trickle charger. Instead, 10-20mA for 750mAh cells is plenty here. Given that, is your objection valid? I know I haven't noticed the cells getting really hot at the end of the charge cycle. They barely get warm in my experience.

It depends on the internal resistance of the battery, the higher it is the bigger the error. Ri when the battery is used as a powersource is very low, in the mOhm. I can't find any data on Ri during charging but considering the batteries I have get warm even on a 0.1C tricklecharge I must assume that it is considerably higher.

Another advantage of charging through the regulator is that you get a constant current. Damaged or depleted batteries might see most of the 24V most people will use. 1kOhm at 24V is .576W. At full charge the Voltagedrop over the resistor is only 24-0.7-1.2x18 or 1.5, making the tricklecharge 1.5 mA.

Quote:

Want to draw this for me? I'm not sure what you mean. Also, be sure you account for the fact that Vce will vary based on current through the transistor. I'm afraid you're going to nearly reinvent the TLE2426 in discretes. What would be the point in that?

Vce will be close to Zero if you saturate the transistor. Assume Vcc is 18V and you want to use a 20mA LED, with a hfe of around 200 on the transistor this translates in to a 0.1mA current over the base, double that to make sure it is open and you get 18V-0.7/0.2mA=86.5KOhm.

Math taken straight from my electronics textbook and I'm using a variation of the circuit in another relay control circuit.

Replacing 2 railsplitters with a few resistors and 3 transistors is a nice saving, that's the point

Quote:

The TLE2426es are only needed when the supply voltage is over 20V. Surely 10V is enough to power this chip reliably?

Absolutely, but a Zener would stabilize Vcc. The closer to 18V you operate the mc33340 the more tolerant to Voltage fluctuations it would be because an error of say 0.1V+ is 2.4V at 24V. Vcc changes 1.2V in this case. At 12V this is 10%, at 18V that is 7%

 

[gracky]

It seems that the noise problem was by multiple reasons, as Tangent said the clicking noise was dealt with paralleling the battery board, and there was some hum like noise and I suppose it was from wall power, I've built a 15v regulated (lm317) power today and it causes no noise.

ppl : My amp is very similar to Toni Kemhagen's one ( http://headwize2.powerpill.org/proje...mhagen_prj.htm ) except that I use Walt Jung's diamond buffer ( http://www.elecdesign.com/Globals/Pl...tent/2800.html ) than IC buffers and I adopted some features of PPA. (FET power rail, TLE2426 virtual ground, biasing opamps to class A...).

Another question, if battery board is paralleled to the wall power, it seems that Vin to bat board and Vout from the board are shorted via wall power, causing that when wall power is off, current from battery goes back to the battery. Now I simply cut off Vin when power is on using dpdt switch (giving up switching minus power rail). Any better idea?

A minor problem remains, I've optimized Jung buffer to battery voltage (about 10v, R9 to 8k). At 15v of wallpower, sound is dumb and dynamic range is too narrow (too much idle current I think) -_-;. I'm considering switching between 2 configurations of the power.

Thank you evryone helped.

 

[tangent]

Quote:

Math taken straight from my electronics textbook

I understand the math, I do not understand what circuit it applies to. Please draw it.

Quote:

current from battery goes back to the battery.

Please examine the PPA amp board and battery board schematics more closely. The input and output diodes you see are arranged to prevent this from happening.

 

[gracky]

Quote:

Originally posted by tangent Please examine the PPA amp board and battery board schematics more closely. The input and output diodes you see are arranged to prevent this from happening.

I see now, I was confused because my layout is different than PPA slightly. Thanks!

 

[Chipko]

Did some thinking and some studying and have managed to get rid of the relays and except for the regulator and charging IC the circuit is now completely discrete.

Charger rev 2


Q3 and Q4 needs to handle the charging current, choose transistor that can handle the appropriate Ic.

D3 and D5 are only needed if the supply voltage is higher than 18V. Choose them so that Vz+Vcc is at all times higher than the supply voltage.

R1, R10= hfe*((V+-Vz)/2,5*Ic) Where Ic is the max charging current

R3, R9=10*R1

R2= Vcc*hfeQ2*hfeQ4/2*Ic

Trickle charge current is calculated as 1.25/RT

Fast charge current is 1.25/RT//RF = 1.25*(RT+RF)/(RF*RT)