class ab questions

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
so i have a class ab setup with one pair of output devieces
and they run really hot with no input
i am using to diodes with a 5k resistor from the +50 voltrail to the two diodes
am i over biasiing the diodes and this is why my output pair is running really hot?
do i need and extra set of output pairs
i am using +- 50 volt rails as you can see
also my output devices are
mjw21193 and mjw21194

why are the devices runnign so hot and by hot i mean i cant touch the heatsink because it will burn my bad
i have the diodes on the heatsink also to prevent thermal runaway

and info insight wioould be greatly appreciated
also one last question
i am wondering where everyone one gets there rca and binding post terminals for the final projects
?
thanks
 
here is my schematic
the output devices are mjw21193/94 from on semiconductor
the diodes are n4001 eventhoug the picture states different
the emmitter resistors are .33 5 watt
and the other resistor is currently 6k
now what do you think?
need more info?
my rails are +- 50
 

Attachments

  • circuit1.jpg
    circuit1.jpg
    8.6 KB · Views: 360
Use a couple of 1N4148 if you want to stick with diodes.

Alternatively, use one diode only and a variable resistor in series. Then you can set the quiescent to whatever you require. Make damn sure that the diode is sitting atop one of the power transistors to prevent thermal runaway.

Even then I'd suggest a Vbe multiplier using a BD139 might be more effective.

Cheers,

Hugh
 
Hi roofingboom,

From your first post it was not clear to me what you were trying to achieve. But sreten is right, the whole circuit will not do what I think you want.
Looking at the +/-50V supplies you want to make an output amplifier able to deliver some 100W in 8Ohm. That means you have to be able to source and sink 4..5A to and from the output. Your output transistors, not being darlingtons, have a current gain around 50. That means that the base current can be as big as 100mA. If your load is resistive that output current must be delivered by the NPN transistor when the amplifier is delivering a high positive voltage (or sinked by the PNP for a negative output voltage). At that moment the voltage across your R3 of 6k will not be much and R3 will only be able to supply a few mA of base current. Certainly not the 100mA that was required.
So, as sreten said, you need drivers in front of the output transistors. You could make a darlington or a compound transistor (CFP). This will increase the current gain and maybe R3 will be able to deliver the required base current. Only for positive output signals this will still be difficult for R3, because the current R3 can deliver will be smaller for higher output voltages. So just when you need the most base current R3 cannot deliver. For that reason you could use a bootstrap circuit or a constant current source instead of R3.
If you choose for an NPN and a PNP darlington output, your bias voltage generator should have a voltage of around 4Vbe or 2.8V. Furthermore this voltage should be temperature dependent and less for increasing temperature. Otherwise the idle current in the output transistors will rise with temperature and eventually get overheated (thermal runaway). The easiest way to make such a bias voltage is to use a Vbe-multiplier, being just a transistor and 2 resistors. The transistor should be thermally coupled to the output transistors.
Have a look at all the circuits that have been posted on this forum and try to understand the functions of all components. Read a book about power amplifiers, e.g. Self on Audio (Self) or Audio Power Amplifiers (Duncan).

Steven
 
i posted a new schmenatic
is this called bootstrapping?
what is the principle behind it and what does it do?
how does it work mathematically and in theory?

can i replace the 2 resistors and capactiro with a current source?
what is better using a current source or bootstrapping?
i am going to eventually going to place a Vbe multipier in the circuit also
i changed the diodes to a different kind and my output transistors are still heating up really bad
 

Attachments

  • circuit2.jpg
    circuit2.jpg
    12 KB · Views: 266
here is full schmetiac i pieced togehter from my readings
the input is via a cap and resistor to ground making a high pass at 6-7 Hz
i am wokring on making the output stage a darlington
my out[put stage is ghastly overheating
i am still wondering what is worng
emiiter resistor for LTP is 22k
emtter resistor for the driver is 33
the resistors for otuput stage are .33
any help would be great
thanks
 

Attachments

  • circuit3.jpg
    circuit3.jpg
    27 KB · Views: 288
sreten said:
As I suspected your circuit is almost certainly oscillating
at high frequencies causing the ouput stage heating.

http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm


Hi !

I can definitely confirm that problem, i had heavy oscillation with
my actual project, the resulting quiscent current jumped from 100ma
to 1.2 A ! (According to spice it did oscillate with 16mhz... ouch)

A year ago i experimented with diodes for biasing, i'll never do it
again, i blowed the transistors and my speakers, as the voltage
generated from rectifierdiodes seems to be much higher than the
Vbe voltage needed for the biasing...

I'm a little confused with your above mentioned link, it says FETs in
the Inputstage are a bad idea ? :scratch:
In my latest design i used jFets the first time for the diffamp, and
this amp completely outperforms any circuits i've built before !
I've never heard before that much details,controlled bass and
crystalclear trebles...
What does that mean ?
That jfets dont produce low THD, but might sound better ?

Michael
 
MikeB said:

I'm a little confused with your above mentioned link, it says FETs in
the Inputstage are a bad idea ? :scratch:
In my latest design i used jFets the first time for the diffamp, and
this amp completely outperforms any circuits i've built before !
I've never heard before that much details,controlled bass and
crystalclear trebles...
What does that mean ?
That jfets dont produce low THD, but might sound better ?

Michael

D.Self doesn't entertain the idea that device types have magical properties.

He gives FETs a thorough bashing as output devices (in aB amplifiers),
and simply maintains an emitter degenerated BJT (to ~ the same gm
as a FET) is simply more linear, I can't disagree with him, as he's right.

:) sreten.
 
The bottom connection of R5 should be at -ve rail, along with the emitter resistor of voltage amplifier.

Because you have no drivers, the current output of the amplifier will be greatly restricted. I'd suggest with say 8mA of VAS current (roughly rail voltage/sum of two boostrap resistors) you could drive high beta, low current output transistors to about 1A peak current.

This is only about 15Vpp, which corresponds to 3.5W into 8R. This makes for a great headphone amplifier, though you could drive a Lowther or Fostex...... :clown:

You could make this into a decent power amplifier with power darlingtons in the output stage; this would work reasonably well but darlingtons give no access to the emitters of the drivers, important for ths type of Double Emitter Follower output stage.

So, you need another cascaded set of devices for the output stage. This is where mosfets have an advantage.....;)

Suggested are 2SC4793/5200 (dvr/OP) and 2SA1837/1943. These are very good, inexpensive Toshibas rated over 30MHz, 230V, 12A, 125W. Beta in both cases is nominally over 50. I use them in all my amplifiers, and they are robust, fast, linear and economical.

Then you will need to consider stabilisation regimes for the amplifier. Will you use a lag compensation cap from collector to base on the voltage amplifier, typically 100pF, and will you also add phase lead from the collector of the VAS to the feedback node (typically 22pF).

Stabilising the amplifier is crucial to performance. It makes the difference between sound and music.

Cheers,

Hugh
 
roofingboom said:
so i have a class ab setup with one pair of output devieces
and they run really hot with no input
i am using to diodes with a 5k resistor from the +50 voltrail to the two diodes
am i over biasiing the diodes and this is why my output pair is running really hot?
do i need and extra set of output pairs
i am using +- 50 volt rails as you can see
also my output devices are
mjw21193 and mjw21194

why are the devices runnign so hot and by hot i mean i cant touch the heatsink because it will burn my bad
i have the diodes on the heatsink also to prevent thermal runaway

and info insight wioould be greatly appreciated
also one last question
i am wondering where everyone one gets there rca and binding post terminals for the final projects
?
thanks


Try this... sorry for the poor quality, I drew it quickly...
 

Attachments

  • output.zip
    4.7 KB · Views: 67
Re: Re: class ab questions

cunningham said:



Try this... sorry for the poor quality, I drew it quickly...


AAHHHHH roofington, my fellow American, let me help you a little...
:cool:
In the realm of BJT's: If you want to make a CLEAN direct coupled output stage that can handle a lower impeadance it should be an emitter follower stage [or voltage follower]. (consult the AC model of a BJT as to why). 8 OHMs is not a really that low of impeadance. Sure it is true that you may loose some volts on your AC but the signal impeadance is lowered dramatically, depending on the beta of your outputs.
If you use darlington package (integrated) output drivers, you don't need a lot of driver current and can use smaller devices reducing noise. BuuutUUUUt, you can run into temp. coefficent problems and thermal runaway because temp. coefficent is related to beta (Hfe, for you beginners) and is large in darlington type devices. (B1 * B2) One mistake that is commonly made is using darlington devices and trying foolishly to bias these with 4 diodes mounted with outputs and this just will not do. As far as temp. coefficents go, you are trying to say [B1*B2=B1+B2] you cannot match beta this precisly in the real world. If you use that circuit, tis is bad, unless you like buying lots of output drivers. BJT's take less to turn on as they are heated(+T coefficient) and MOSFET's take more voltage to turn on as they are heated.(-T coefficent)Which can drive them to crossover distortion. Don't ever place two transistors in darlingon on class AB circuits and expect stable operation However, there is a way.
What if you want to drive 4Ohms. Best to leave Mr. Darlington with voltage and current regulator circuits for the sake of stability. This means that if you want real liniear current output than you need high current drivers. They usually have a beta around 30-50. 100WRMS is 5A through 4Ohms.
5A/30=160mA The driver stage must output 160mA RMS to run the outputs. (always underestimate) Since Z out is 4Ohms, then
4*30=120 So make it drive 100Ohms. (always over-rate) Driver stage must be able to handle 200mA through 100Ohms. 120mA * 100Ohms is about 20V RMS White noise is related to heat. I suppose as electrons slam into atoms of resistors and semiconductors they make noise, unavoidable. White noise is very small and usually unnoticeable, but it can be amplified though. {so filter it}
hissssssssssssssssssssssssssssssssssssssssssssssssssssssssssssssss

So if you use the drivers to bias the outputs, you should add small emitter resistors because you want to make a stable circuit that you can leave on and go to the store and not worry that your pad done burnt down. Doing this will make the bias of the circuit NOT Beta dependent. (consult the AC model of a BJT as to why). This means that if you replace one of the transistor with one of same type, you don't really have to worry about who made the device. Yes you loose some gain, but your wallet will thank you for it in the long run.
Using a bias servo transistor(ECG185) is the best way to maintain the proper bias current through the bi-polar push pull outputs. This device must mount with outputs and have similar Beta value. Now you can even adjust the bais current. If you do as I have done in this drawing, you should know that the driver stage must output 3-5 Watts into the output transistors to run them at 5A RMS. Use a small seperate heatsink 5W or so and the temp. coefficent of this device will not be a bias problem.
In the drawing, D1 & D2 are power rectifier diodes. they have a larger PN voltage than outputs do, but they are biased by the servo. This acts kind of like a safety catch in case your wiring is s%$tty. A zener voltage regulator and a DC amp to give a steady +50V -50V and a bias current limiter (1.8mA) will keep the noise out of your ears.
The diff. amp, VAS, and DC offset circuits I will let you figure it out because I'm not GIVING all of my work away, but I hate to hear people struggle and waste expensive outputs. If you make a feedback circut that can drive 24V RMS into 5KOhms, then you can direct couple input and output with feedback connection from output. All resistors in this circut can be 1/8 Watt unless noted.
My diff., VAS, DC offset circuits all use 1/8 Watt resistors except for four. It also uses dual pwr supplies.

I value my hearing... You can always adapt this circuit to a smaller circuit. It may be more complicated than min requirements, but if constructed properly will not smoke when you are away.

Anyway, I hope this my point you in the right direction with BJT's. bye :
 

Attachments

  • bjt_current_amp.jpg
    bjt_current_amp.jpg
    74.5 KB · Views: 183
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.