Re: Transistor Stereo Amp


Jan 30, 2011

 


----------------------------

#362 Jan 30, 2011

I'm jumping in the silicon Corvette again. Today I made a power supply for my amplifier, which I intend to modify shortly. It's a 12 VAC transformer at 450 mA, a fullwave bridge, and a pie filter made of two 1000 uF caps and a 39-ohm resistor.



I think you know what I intend to change. At modest current levels the filter resistor drops too much voltage, which throws off supply voltages and can unbias transistors. This is one of my main complaints with transistors. I intend to swap it with a 10 ohm or less resistor, one that won't steal my precious voltage. Maybe I should ditch the resistor, add more filters, and decouple each stage individually.



I haven't set a CCS in the bias network of the power transistors yet. I don't know how to calculate voltage and current for the networks but I intend to learn somehow. I'm staying away from rubber diodes right now. I'm still having the same problem: the driver transistor clips much sooner than I need it to. I thought about setting a third series diode in the biasing network, but that seems not to have any effect.



This one did get me a whole watt of output power, although it did make my sine wave look like legos. I'd like to know how to make the output sweep across the entire DC range. My amps seem to clip long before the entire range (less than 50%) and I don't get a theoretical full sweep. All of the transistor tutorials I've read tell me I can get the output to range from DC to 0v. What gives?



Ed



----------------------------

#363 Jan 30, 2011

Hi Ed,



A couple of ideas: 1st, why not just simply get rid of the 39-ohm resistor and put a series-pass regulator of the appropriate voltage in its place? (Unless of course, you just absolutely need all that voltage) You will need to find out what the actual operating voltage the circuit needs and design around those parameters.



2nd, considering most SS amp designs, a regulated power supply is not usually necessary, as the amp tends to "act" like a regulator per se. (It's not worded exactly right, but that's the behaviour of the amp anyway.)



Are ou making an amp from another design, or are you trying to design one yourself?





Bob, KD5MHQ

--- On Sat, 1/29/11, Der_Hellste_Stern lordof16bit@...> wrote:



> I'm jumping in the silicon Corvette

> again. Today I made a power supply for my amplifier, which I

> intend to modify shortly. It's a 12 VAC transformer at 450

> mA, a fullwave bridge, and a pie filter made of two 1000 uF

> caps and a 39-ohm resistor.

>

> I think you know what I intend to change. At modest current

> levels the filter resistor drops too much voltage, which

> throws off supply voltages and can unbias transistors. This

> is one of my main complaints with transistors. I intend to

> swap it with a 10 ohm or less resistor, one that won't steal

> my precious voltage. Maybe I should ditch the resistor, add

> more filters, and decouple each stage individually.

>

> I haven't set a CCS in the bias network of the power

> transistors yet. I don't know how to calculate voltage and

> current for the networks but I intend to learn somehow. I'm

> staying away from rubber diodes right now. I'm still having

> the same problem: the driver transistor clips much sooner

> than I need it to. I thought about setting a third series

> diode in the biasing network, but that seems not to have any

> effect.

>

> This one did get me a whole watt of output power, although

> it did make my sine wave look like legos. I'd like to know

> how to make the output sweep across the entire DC range. My

> amps seem to clip long before the entire range (less than

> 50%) and I don't get a theoretical full sweep. All of the

> transistor tutorials I've read tell me I can get the output

> to range from DC to 0v. What gives?

>

> Ed



----------------------------

#364 Jan 30, 2011

Hello Bob,



I am trying to design this amp myself. I don't know all that I want to about BJTs, therefore I intend to tackle them full-force. I'd like to design it in the same fashion as one would design a tube amplifier, but that's going to be somewhat impossible. Transistors are just too different. They are, however, much cheaper than tubes and I have quite a lot of them.



I wouldn't think that the circuit acts like a regulator; a class AB amplifier will draw more current with stronger signal input. This causes the power supply voltage to sag more, which somewhat debiases the transistors. That 39 ohm resistor helps this; if the circuit were to draw 100 mA the power supply would sag, and the resistor drops around 5v, which takes my total down to 10. I get an unloaded voltage of 18.8, and I want to use as much of that as I can.



I don't intend to put a series regulator on the P/S board, as the board is already made and there isn't room for one. I could certainly make another board just for a regulator, but I'd want the voltage to be high as I could get it. When running preliminary tests on this circuit the B+ I used was 12 and some change, but I'd hoped that a higher voltage would give me more headroom. As stated before, I always have problems getting any sort of amplifier to swing fully between ground and B+. All of the transistor tutorials I've ever read claim that you can get a "full signal swing" but I can't seem to find that. Best I find is about 40-50%, and then clipping happens. That's on small signal stages. For class AB output I might get 10% of the DC range as a usable swing.



I do know that I'm replacing the 39 ohm resistor with one of lesser value, in order to save my voltage. I'll probably use 10 ohms. It doesn't do very much for filtering anyhow, I'd need higher value resistors and lower levels of current to make for a quieter supply.



Ed







----------------------------

#365 Jan 31, 2011

Hi Ed,



I think it's great that you're trying to design an amp as you're doing. I'd suggest just building one from a tried'n proven schematic. That's what I've done in the past and have had great results.



I can understand that BJT's can be a bit difficult to understand, espcially when you've been working with valves and then making the transition to transistors. Yup, transistors behave differently than do valves. But they still need to be biased and in such a way to make them operate in the class needed...just like valves.



When I said that the circuit "acts" like a regulator, you noticed the way I stated that, right? It was stated in such a way to convey how these audio circuits work. A regulated power supply is seldom ever needed for such audio amps, due to the way the circuit behaves. Think of a tube-type linear amplifier- they don't have a regulated supply, yet they work just fine. Due to limited space here, I won't explain the whole theory of operation for these electronic principles, other than, they tend to "self regulate" (notice that I put that phrase in quotes), due to how the circuit is designed.



What kind of output power do you want from your amp? I have several designs that you may be interested in.



Bob, KD5MHQ

--- On Sun, 1/30/11, Der_Hellste_Stern lordof16bit@...> wrote:



> Hello Bob,

>

> I am trying to design this amp myself. I don't know all

> that I want to about BJTs, therefore I intend to tackle them

> full-force. I'd like to design it in the same fashion as one

> would design a tube amplifier, but that's going to be

> somewhat impossible. Transistors are just too different.

> They are, however, much cheaper than tubes and I have quite

> a lot of them.

>

> I wouldn't think that the circuit acts like a regulator; a

> class AB amplifier will draw more current with stronger

> signal input. This causes the power supply voltage to sag

> more, which somewhat debiases the transistors. That 39 ohm

> resistor helps this; if the circuit were to draw 100 mA the

> power supply would sag, and the resistor drops around 5v,

> which takes my total down to 10. I get an unloaded voltage

> of 18.8, and I want to use as much of that as I can.

>

> I don't intend to put a series regulator on the P/S board,

> as the board is already made and there isn't room for one. I

> could certainly make another board just for a regulator, but

> I'd want the voltage to be high as I could get it. When

> running preliminary tests on this circuit the B+ I used was

> 12 and some change, but I'd hoped that a higher voltage

> would give me more headroom. As stated before, I always have

> problems getting any sort of amplifier to swing fully

> between ground and B+. All of the transistor tutorials I've

> ever read claim that you can get a "full signal swing" but I

> can't seem to find that. Best I find is about 40-50%, and

> then clipping happens. That's on small signal stages. For

> class AB output I might get 10% of the DC range as a usable

> swing.

>

> I do know that I'm replacing the 39 ohm resistor with one

> of lesser value, in order to save my voltage. I'll probably

> use 10 ohms. It doesn't do very much for filtering anyhow,

> I'd need higher value resistors and lower levels of current

> to make for a quieter supply.

>

> Ed



----------------------------

#367 Jan 31, 2011

Hello Bob,



I kinda am using tried-and-true schematics, and I'm kinda not. At first I was trying to figure out how to get an emitter follower to drive a speaker. Dog no hunt. I want to find an amp circuit but make it my own so I learn how it works. Building kits doesn't do much for the educational process.



I've had to look far and wide to find ideas on how to do this, as making an amplifier that has a > 100K input impedance and .1R output impedance is a challenge indeed. Normally when I design CE amps I end up with less Zin than Zout. I keep having to look at tutorials and schematics to get a clue what I am doing.



I am trying to make a transistor amplifier that can output 5w into 8 ohm speakers. I want my amp's acceptable volume level to be lower than maximum. In other words, "too loud" would be at fully CW and comfortable listening would be about halfway. This way I have some power reserve.



My transformer is a 12.6 VCT, 450 mA, hooked up to a FWB and a pi filter. The capacitors are 35v units separated by a 2.2 ohm resistor that is not much more than a fuse now. The power supply sags more than I'd like but the only way I can think to control this is add much more capacitance. I can't put a high series resistance in the main filter because it would eat up all of my voltage. The voltage drops to 16 or 17 volts under load.



i1094.photobucket.com/albums/i443/Der_Hellste_Stern/TransistorAmp.gif?t=1296497229



Here again is my schematic. I put a 1 ohm resistor in the emitter branch of each power transistor. Ignore the CE stage for now; the only important part is everything to the right of the second coupling cap. Something I've noticed is where the waveform clips; clipping starts at the collector of the drive transistor, i.e. the one below the two diodes (in that schematic it's a rubber diode instead of two diodes), at outputs higher than 2vpp. I can get a higher output but it will have so much distortion it's useless. I have tried playing with the value of the 2.7K resistor to no avail. One thing I do have a problem with is my iPod's maximum voltage output. If I try upping the sine wave to more than 1vPP the signal starts to clip. Thus, I can't tell if the amp is clipping unless I look at both input and output.



What I need to build is a small gen-purp amplifier with an extremely low impedance. It will need to be able to drive loads as low as 10 ohms and have a gain of ~5. It will go into an Altoid's can and be another test tool at my disposal. If I give it a ridiculous amount of gain then maybe I can NFB it into low distortion. Perhaps I need to do that instead of spinning my wheels blindly with this.



I do intend to drive the amplifier directly, so it will have its own preamp stage. This can-amp will just be a necessary tool to drive low impedances. If I had a sine wave generator that was worth anything I would use it instead. Now I have to figure out how to make a transistor amp with a Zin of 100K and a Zout of 1 ohm. Fun?



Now a comment on the "self-regulation" you spoke of. A tube amp normally works with high voltage. Say, a B+ unloaded of 250. Now our 6C4 starts to conduct and B+ drops to 240, ten volts. This percentage of change is so small that not much is affected. Second, the plate resistance of triodes will be fairly high and the operating point does not change much.



With BJTs, your voltage is much lower, so a ten-volt sag on an 18-volt supply will have much more an effect. Even if the voltage change would be proportional to the change in the first example, say .4 volts, this affects the bias networks. Tube bias networks are not connected to B+ like transistors are. If using voltage divider bias, the base voltage will drop as the B+ does, dropping the emitter voltage and lowering current. As current lowers, the Vc goes up whereas the B+ went down, moving the Q point closer to the top edge of the voltage rail. If one is trying to get a full swing from rail to rail this occurence will cause clipping, even if the amp had no distortion with a regulated power supply. With tube circuits, I don't recall ever trying, or being able, to get output swing from ground to B+. With a transistor circuit I'd want to use 18v if I wanted to get 9vpp on the output. Seems to me that distortion would decrease.



I don't know enough about transistors. I think there's some magic secret that I haven't learned yet. I'll look at your designs in hope that I can gain some wisdom.



Ed







----------------------------

#371 Jan 31, 2011

Well, your circuit looks vary good--very close to a typical bipolar xstr design. I applaud you for wanting to learn how these SS amps work.



There's a few things I think you may need to check...



* What method is being used for determining output power?

* Is a resistive 8 ohm load being used for the test?

* What is the test frequency?

* What is the rail voltage at maximum output (at clipping)?

* What is the idle current (measured at TIP31C collector)?

* What is the purpose for the 1-ohm resistor?



There is one possible problem I noticed: The power transformer. It appears that your power transformer may be inadequite. It should be at least a 1 amp rating. Remember, these amps are usually only about 60% effiecient, NOT 100%. In other words, using the data you stated, your power transformer is rated at 12.6 VAC (RMS) at 450 mA. 12.6 V x 0.45 A = 5.67 watts DC power. Your amp will need at least twice that DC power to effectively deliver the 5 watts you want at the load.



I'd like to suggest that you do the following:



* Test your amp on a 12v regulated power supply of at least 1.5 amps;

* Replace 330 uF output capacitor with at least a 1000 uF

* Try testing the amp with a 1 kHz sinewave with output into an 8 ohm

resistive load (like a Radio Shack 8 ohm 20 watt resistor Catalog #

271-120). These resistors make great dummy loads. See this web site:



www.radioshack.com/product/index.jsp?productId=2062288



Try what I mentioned here first. In this manner, you can properly find out what's going on in your circuit, then you can more easily figure out what the problem is. I see no reason this circuit can't provide the output power you want. Lemme know what you find.



Bob

--- On Mon, 1/31/11, Der_Hellste_Stern lordof16bit@...> wrote:



> Hello Bob,

>

> I kinda am using tried-and-true schematics, and I'm kinda

> not. At first I was trying to figure out how to get an

> emitter follower to drive a speaker. Dog no hunt. I want to

> find an amp circuit but make it my own so I learn how it

> works. Building kits doesn't do much for the educational

> process.

>

> I've had to look far and wide to find ideas on how to do

> this, as making an amplifier that has a > 100K input

> impedance and .1R output impedance is a challenge

> indeed. Normally when I design CE amps I end up with less

> Zin than Zout. I keep having to look at tutorials and

> schematics to get a clue what I am doing.

>

> I am trying to make a transistor amplifier that can output

> 5w into 8 ohm speakers. I want my amp's acceptable volume

> level to be lower than maximum. In other words, "too loud"

> would be at fully CW and comfortable listening would be

> about halfway. This way I have some power reserve.

>

> My transformer is a 12.6 VCT, 450 mA, hooked up to a FWB

> and a pi filter. The capacitors are 35v units separated by a

> 2.2 ohm resistor that is not much more than a fuse now. The

> power supply sags more than I'd like but the only way I can

> think to control this is add much more capacitance. I can't

> put a high series resistance in the main filter because it

> would eat up all of my voltage. The voltage drops to 16 or

> 17 volts under load.

>

> i1094.photobucket.com/albums/i443/Der_Hellste_Stern/TransistorAmp.gif?t=1296497229

>

> Here again is my schematic. I put a 1 ohm resistor in the

> emitter branch of each power transistor. Ignore the CE stage

> for now; the only important part is everything to the right

> of the second coupling cap. Something I've noticed is where

> the waveform clips; clipping starts at the collector of the

> drive transistor, i.e. the one below the two diodes (in that

> schematic it's a rubber diode instead of two diodes), at

> outputs higher than 2vpp. I can get a higher output but it

> will have so much distortion it's useless. I have tried

> playing with the value of the 2.7K resistor to no avail. One

> thing I do have a problem with is my iPod's maximum voltage

> output. If I try upping the sine wave to more than 1vPP the

> signal starts to clip. Thus, I can't tell if the amp is

> clipping unless I look at both input and output.

>

> What I need to build is a small gen-purp amplifier with an

> extremely low impedance. It will need to be able to drive

> loads as low as 10 ohms and have a gain of ~5. It will go

> into an Altoid's can and be another test tool at my

> disposal. If I give it a ridiculous amount of gain then

> maybe I can NFB it into low distortion. Perhaps I need to do

> that instead of spinning my wheels blindly with this.

>

> I do intend to drive the amplifier directly, so it will

> have its own preamp stage. This can-amp will just be a

> necessary tool to drive low impedances. If I had a sine wave

> generator that was worth anything I would use it instead.

> Now I have to figure out how to make a transistor amp with a

> Zin of 100K and a Zout of 1 ohm. Fun?

>

> Now a comment on the "self-regulation" you spoke of. A tube

> amp normally works with high voltage. Say, a B+ unloaded of

> 250. Now our 6C4 starts to conduct and B+ drops to 240, ten

> volts. This percentage of change is so small that not much

> is affected. Second, the plate resistance of triodes will be

> fairly high and the operating point does not change much.

>

> With BJTs, your voltage is much lower, so a ten-volt sag on

> an 18-volt supply will have much more an effect. Even if the

> voltage change would be proportional to the change in the

> first example, say .4 volts, this affects the bias networks.

> Tube bias networks are not connected to B+ like transistors

> are. If using voltage divider bias, the base voltage will

> drop as the B+ does, dropping the emitter voltage and

> lowering current. As current lowers, the Vc goes up whereas

> the B+ went down, moving the Q point closer to the top edge

> of the voltage rail. If one is trying to get a full swing

> from rail to rail this occurence will cause clipping, even

> if the amp had no distortion with a regulated power supply.

> With tube circuits, I don't recall ever trying, or being

> able, to get output swing from ground to B+. With a

> transistor circuit I'd want to use 18v if I wanted to get

> 9vpp on the output. Seems to me that distortion would

> decrease.

>

> I don't know enough about transistors. I think there's some

> magic secret that I haven't learned yet. I'll look at your

> designs in hope that I can gain some wisdom.

>

> Ed







----------------------------

#373 Jan 31, 2011

Hi Bob,

> Well, your circuit looks vary good--very close to a typical bipolar xstr design. I applaud you for wanting to learn how these SS amps work.



I think my lack of knowledge is the perfect reason for me to learn. Also there are many advantages to this, such as not having to buy transformers or 300-volt caps.

> There's a few things I think you may need to check...

>

> * What method is being used for determining output power?

Measurement of P-P voltage across 8 ohm resistor.

> * Is a resistive 8 ohm load being used for the test?

Yes.

> * What is the test frequency?

Normally 1K but I play around with different ones. Also I use square and triangle waves too.

> * What is the rail voltage at maximum output (at clipping)?

Not measured yet. When I do the DC measurements it normally ends up around 17-ish volts.

> * What is the idle current (measured at TIP31C collector)?

Varies based on how I configure my circuit, which I change constantly to see the effects of various things.

> * What is the purpose for the 1-ohm resistor?

Measuring the idle current of the power transistors. They also provide a minute amount of DC feedback if the output devices decide to go into thermal runaway.

>

> There is one possible problem I noticed: The power transformer. It appears that your power transformer may be inadequite. It should be at least a 1 amp rating. Remember, these amps are usually only about 60% effiecient, NOT 100%. In other words, using the data you stated, your power transformer is rated at 12.6 VAC (RMS) at 450 mA. 12.6 V x 0.45 A = 5.67 watts DC power. Your amp will need at least twice that DC power to effectively deliver the 5 watts you want at the load.



Doesn't seem like a hard task to replace the transformer. All I'd have to do is replace a couple of wires. I don't see myself getting 5 watts in one channel, and I'm intending to make two.

> I'd like to suggest that you do the following:

>

> * Test your amp on a 12v regulated power supply of at least 1.5 amps;

> * Replace 330 uF output capacitor with at least a 1000 uF

> * Try testing the amp with a 1 kHz sinewave with output into an 8 ohm

> resistive load (like a Radio Shack 8 ohm 20 watt resistor Catalog #

> 271-120). These resistors make great dummy loads. See this web site:

>

> www.radioshack.com/product/index.jsp?productId=2062288

>



Can do what you recommended. I have two of those resistors. Soon I will need to formulate somehow a whole bunch of them to make a 200W, 8-ohm resistor. That way I can build Amp-Zilla. Then I will buy a Flying V guitar and elect the dead. I have a computer PSU that will suffice for the test supply, and I know I have that large an output capacitor somewhah.



I loaded my iPod with the various test signals I use. It seemed like the best way to simulate real-operation conditions. As mentioned earlier I do intend on making a small amplifier so I can simulate much larger voltage swings. Maybe I can make it go as high as 6 or 7 vPP.

> Try what I mentioned here first. In this manner, you can properly find out what's going on in your circuit, then you can more easily figure out what the problem is. I see no reason this circuit can't provide the output power you want. Lemme know what you find.

>

> Bob



I don't necessarily need 5 WPC here but I'd be satisfied w/ two or three. I may not get to toying with this circuit further until I get back to Washington. There are several things I have to do before we can move, so electronics might take the back burner here shortly.



Ed



This Site's Privacy Policy
Google's privacy policies

S
e
n
i
o
r
T
u
b
e
.
o
r
g