IMO the two hardest things to control in a simple amp are biasing and DC-offset voltage.
Some books teach complete garbage when it comes to amps, such as biasing amps with two diodes, instead of VBE multiplier, and explaining how to set up common emitter or base voltage amp, but not really explaining how they go together with output stages, or how you use the diff stage to control the voltage stage, not just feed it a signal.
A good working amp design uses 3 stages: Differential amp, voltage gain (common emitter) and output stage (two emitter followers joined together)
A diff stage makes the amp easier to build as you don't have to guess resistors all day to get a good DC-offset, because the diff stage does it for you. Also to make amp building easier, for biasing the output stage, a simple 1 transistor and 1 trimmer POT in the diodes place is MUCH better than two diodes and having to guess bias resistors.
Also check out Rod Elliott's site with several designs:
www.sound.westhost.com
Some books teach complete garbage when it comes to amps, such as biasing amps with two diodes, instead of VBE multiplier, and explaining how to set up common emitter or base voltage amp, but not really explaining how they go together with output stages, or how you use the diff stage to control the voltage stage, not just feed it a signal.
A good working amp design uses 3 stages: Differential amp, voltage gain (common emitter) and output stage (two emitter followers joined together)
A diff stage makes the amp easier to build as you don't have to guess resistors all day to get a good DC-offset, because the diff stage does it for you. Also to make amp building easier, for biasing the output stage, a simple 1 transistor and 1 trimmer POT in the diodes place is MUCH better than two diodes and having to guess bias resistors.
Also check out Rod Elliott's site with several designs:
www.sound.westhost.com
Hi again.
Well, i was a bit tired writing and I guess it showed...probably no-one understod what I ment, exept myself
Lets try again.
Say that the 405 can be divided into 4 blocks:
-input buffer ic that also is part of a dc servo loop
-an class A amp
-the current dumper
-the current dumping "bridge" ( feedback/feedforward)
The class A amp is the low distorsion amp that corrects errors in the current dumper, and as far as I understand sets the distorsion for the whole amp. Also the feeedback/feedforward for ac is not conected through the input buffer ic.
Now, why hasn't the class A amp a diff input, which ought to give less distorsion?
The only thing that is special for the classs A amp is that it has to have very wide bw to be able to correct for the crossover distorsion from the current dumper but I don't see why that would be a problem with a diff input.
Well, i was a bit tired writing and I guess it showed...probably no-one understod what I ment, exept myself
Lets try again.
Say that the 405 can be divided into 4 blocks:
-input buffer ic that also is part of a dc servo loop
-an class A amp
-the current dumper
-the current dumping "bridge" ( feedback/feedforward)
The class A amp is the low distorsion amp that corrects errors in the current dumper, and as far as I understand sets the distorsion for the whole amp. Also the feeedback/feedforward for ac is not conected through the input buffer ic.
Now, why hasn't the class A amp a diff input, which ought to give less distorsion?
The only thing that is special for the classs A amp is that it has to have very wide bw to be able to correct for the crossover distorsion from the current dumper but I don't see why that would be a problem with a diff input.
bear said:
Are you *sure* you're a 4th year EE student?
This is a subject which I have quite a bit of interest in. The implication here is that any 4th year EE student (and therefore close to being a graduated REAL EE) should know this already. Therefore, you are doubting his qualification to be an EE based on his lack of analog design knowledge in this area. This is a question which comes up rather frequently all over in places where large groups of knowledgeable and experienced EE come together to talk about EE matters. The actual subjects don't matter because in any subject there are always the experienced ones and the ones who are just learning.
In this case, it's a knowledgeable EE or very experienced hobbiest essentially badmouthing a up and coming engineer for his lack of analog knowledge. What is often forgotten is that the world has expanded now. No longer is the only subject of interest analog design methods. In a modern university, analog design, digital discrete design, embedded programming, control systems, programming in high level languages and using simulation programs are all subjects that need to be taught to an incoming EE student. Understandably this does not always leave time for detailed discussions of the many facets of each of these fields.
Expressing public doubt at a students inability to understand a subject is rediculous enough, but it is intolerable when you also doubt his qualifications as an EE because of it. How would you feel if I were to say that you arn't really an EE because you don't know how to write code? Or design an embedded system? First, you'd probably say that it isn't required of you to know such things. And you're completely correct. We live in a wide world and work in a broad field. Remember that not everyone has the same interests as you, and instead of making snide comments about someones qualifications, try to help them. This is a forum for advising people and sharing information, not insults. And don't just keep it on this forum, expand it into every facet of your professional life. Remember that just because he can't figure out why a differential stage is so great, he probably can do atleast one EE thing you can't, but that doesn't make him any better an EE.
please take a look at this amp. I know it's not good , but it's my first one, I'm not thrying to be the best, but just to make it work. Against multisim 7, for the 1mv signal I'm getting a 1.9V peak-to-peak output for the 4Ohm resistor load with zero DC level. Seems good, but the problem is a little distortion on the output signal: its lowest value is - 0.75V and the highest is about +1.2V. Therefore, the output is not a valid sine wave, since it's not symmetrical with respect to time axis. I know I must use negative feedback to correct this, so I want suggestions about what to do to correct this amp using feedback. Please take a look and try to teach me.
Attachments
ashade said:
4 of them will not be able to do it either, not in the real world. In fact, looking at your schematic, it has about the same chances of working outside of a simulator, as a snowball has in hell (Look up some data on semiconductor parameter tolerances, as well as secondary breakdown - neither of which are modelled ina simulator). Still, since you do not seem to be willing to learn from answers you get here, we will just all let you learn for yourself.
hi ashade,
when looking at your schematic I can see that R7 and Q5 act as a simple resistor load for Q1. Assuming that Q1 and Q5 have a beta of about 150, the maximum current Q1 is able to source is 150 * 2ma = 300ma. No wonder that you can't get more than 1,2V positive voltage swing.
The same with the negative voltage, you build a nice voltage divider. 55V * 4 ohms / (4ohms * 50k / beta) = 0.65V. No need for simulating this.
D1 to D4 in your schematic are useless. They do not bias anything. If they would they'll blow Q1 and Q5. Trying to get your amp running is a waste of time. Sorry to say this, but there is no potential in. You designed this amp by throwing some parts in the simulator and trying to get it running. Probably you're also simulated it wrong, at Tj = 27°C, which will never fit the realitiy for the output devices.
Youl'll better try of one of the JLH designs. They are running and have no differential input. You can simulated it as well as building it. After that try to vary the resistor, capacitor and transistor values / parts. Simulation models never fit the parameter variations of real devices. There is only one parameter which is quite independent from parameter variations: Vbe. But only for the same parts at the same temperature.
when looking at your schematic I can see that R7 and Q5 act as a simple resistor load for Q1. Assuming that Q1 and Q5 have a beta of about 150, the maximum current Q1 is able to source is 150 * 2ma = 300ma. No wonder that you can't get more than 1,2V positive voltage swing.
The same with the negative voltage, you build a nice voltage divider. 55V * 4 ohms / (4ohms * 50k / beta) = 0.65V. No need for simulating this.
D1 to D4 in your schematic are useless. They do not bias anything. If they would they'll blow Q1 and Q5. Trying to get your amp running is a waste of time. Sorry to say this, but there is no potential in. You designed this amp by throwing some parts in the simulator and trying to get it running. Probably you're also simulated it wrong, at Tj = 27°C, which will never fit the realitiy for the output devices.
Youl'll better try of one of the JLH designs. They are running and have no differential input. You can simulated it as well as building it. After that try to vary the resistor, capacitor and transistor values / parts. Simulation models never fit the parameter variations of real devices. There is only one parameter which is quite independent from parameter variations: Vbe. But only for the same parts at the same temperature.
I've changed that schematics to this one below. A temperature sweep test told me it works for a range of about -20 .. 50 ºC. Of course it needs to be improved. Now I'm getting low distortion due to max negative/positive output voltage swing, but it still happens. Also, the capacitors are too large and the frequency response is not good for the bandwith it's supposed to work on (20Hz-20kHz). Now, please see the schematic and post your comments. Be nice this time and remember I'm just a beginner and not a pro like you,
Attachments
There are still major problems with your circuit. All four transistors are biased full on, the first stage will never swing more than 15vP-P so unless you give the second stage some gain the 55v rails are pointless.
I suggest you try simulating some popular amps and play around with modifying them until you are ready to design your own.
I suggest you try simulating some popular amps and play around with modifying them until you are ready to design your own.
you have to put in mind that i'm not making a high power amp, i'm just trying to learn from the most basic circuit possible. That's why I made this circuit. Of course I know the gain must be much higher for real audio amps. The swing distortion is very low on this schematic, I've tested for a 10mV input and almost no distortion. All I ask now is: how do I improve the temperature operation range of the circuit and how do I improve the frequency response of the amplifier for bass signals (300Hz or less) without using larger capacitors.
PS: is there a pnp coupled pair for the mpsa18?? if not, what's the closer PNP to that??
PS: is there a pnp coupled pair for the mpsa18?? if not, what's the closer PNP to that??
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.