diyAudio (
-   Solid State (
-   -   amp (

fetente 13th September 2005 09:51 AM

2 Attachment(s)
HI, my name is Andrea. I am a new user but it's a pair of month I "sail" here. Now I would like to send a post to receive a help.
Attached you'll find a schematic of a simple amp.
If I try it with my function generator, I have a perfect sin wave from 40Hz to 20KHz without any distortion or auto oscillation, but if I try with music from cd output I heard a little fuzzy distortion on mid/hi notes :bigeyes:
Does anyone have a suggestion?

Many thanks,


ilimzn 13th September 2005 10:39 AM

On a typical oscilloscope it is hard to see any distortuion below about 2-3% (sometimes even more), but it will be very audible.
To get a good idea of the distortion products, you should actually compare the output, scaled down by the gain of the amplifier, with the input. Also, have you tried looking small output voltages? Looking at full output sinewaves will hide the distortion because it occurs only around the zero crossings.

Regarding your amplifier, you are asking a lot of the OPA445, since the output stage is biassed more into class C than class B (with BJTs you cannot achieve real class B because they do not have a defined threshold of conduction). A non-obvious consequence is that the amp will produce a lot more distortion on reactive loads.
You may want to include some kind of simple biassing arrangement for the output BJTs.

Your output C needs to be larger to cope with the full audio band. It could also be completely omitted considering you have a +- power supply. Given the OPamp front end, offset should not be a problem. As it is now, it is reverse biassed for negative half-periods, especially at low frequencies, which will be a source of considerable distortion in itself.

Jan Dupont 13th September 2005 04:09 PM

Yes. You will absolutely get zero-crossing distortion, as this amp acts as a class B amp with normal speaker loads.
Insert a bias circuit and remove the output cap ;)

lumanauw 14th September 2005 06:05 AM

Sometime ago Circlotron suggested this type of cct as "current booster". Have you tried bigger output front chip, like LM1875 instead of opamps?

Leolabs 15th September 2005 03:24 PM

U should try "Gainclone" too:)

CBS240 16th September 2005 06:55 AM

2 Attachment(s)
Hi Andrea,

First of all, you should lose the capacitor to the output because the differential amp stage inside the op-amp will correct and zero the DC output and it is not needed.

In order to get a full 360 degrees of sine wave, the emitter follower transistors must be biased class AB. This means that each device must have a small amout(realitively) of DC current flowing through them, just above cuttoff. Otherwise there is a small amount of time between when the NPN cuts off and the PNP turns on, and vice versa where there is no conduction and this is known as dead time, or crossover distortion. Quite noticeable. If you measure the bases of the outputs with a scope as they are in your circuit, you may notice a spike in the sine wave as the op-amp attempts to correct for this. Typically a silicon transistor requires 0.6V between the base and emitter...Vbe. Since you are using TIP142/147, these are darlington devices and this means there are two transistors inside each device. Therefore you have two PN junctions to overcome, and this makes the Vbe turn-on 1.2V. You must have 1.2V between the base and emitter of each device, IOW 2.4V between the bases of the devices in order to get bias.
The best way to bias these devices might be to have a current source on the + rail, and one on the - rail.

One variable regarding transistors is the temperature coefficient. In BJT's it is a positive number. This means that as the transistor gets hot, the Vbe turn-on voltage reduces slightly. So if you have a set voltage to bias them on, as they get hotter, the DC bias gets higher creating more heat, making the bias more and more and so on. The transistors will burn up because the DC load line is very steep for a class AB bias. This is known as thermal runaway and darlingtons are very sensitive to this because the temperature coefficient of the first transistor is multiplied by the coefficient of the second. This is the trade off for such a huge current gain. In order to get thermal stability, you need to have what it called a Vbe multiplier. This creates thermal dynamic biasing so that as the outputs get hotter, the Vbe multiplier is heated and turns on more, reducing the bias voltage on the outputs accordingly.

You would want to choose a darlington transistor for the Vbe multiplier and mount it beside the outputs on the heat sink so they are the same temperature. This can be a smaller device because it does not produce heat, rather absorbs it from the outputs.

In order to set the bias, you want to start out with the pot set to maximum resistance so that the Vbe multiplier is saturated, and the outputs are in cuttoff. Do this with no load connected. Measure the voltage across the emitter resistors of the outputs and it should be 0V. As you adjust the pot, the Vbe multiplier will start to turn off and you will see voltage across the emitter resistors. 50mA might be a decent bias. Use Ohms law, I=V/R to set the amount of current.

I realize this is kind of technical for someone without a lot of experience but I hope it doesn't make you shy to ask questions. It really helps to understand the principles behind operation rather than just copying a circuit, because if you have trouble, you can understand how to fix it and why it is not working properly. I attached an example picture to help explain somewhat.

amplifierguru 16th September 2005 07:32 AM

Hi Andrea,

What many people don't understand is that feedback can only work 'as advertised' when there is a forward path that provides sufficient gain! If you insert a discontinuity such as an un- or under-biassed output stage where gain drops, so does the circuits NFB and ability to correct within it's loop. You cct is one that drops gain significantly so you can't expect NFB to work for you.

If you ensure the forward path maintains it's gain, NFB will work as advertised.

Trust that helps - one little gem I have gleaned in a lifetime of amplifier design.

darkfenriz 16th September 2005 01:58 PM

Yes guru
the amp needs linear transfer and high open-loop bandwidth. Only than feedback will work correctly.
Jusy like a bank credit.
Do you know Murphy's law?
If you want a bank credit you must make the bank believe you actually don't need it.

alphabeta 16th September 2005 02:54 PM

Thanks CBS240 for the detailed explanation. It's helped me a lot understand my Cyrus 2 circuit diagram.

BTW - they use a tripple layer "darlington like" arrangement on the output stage!

All times are GMT. The time now is 06:23 AM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 18.75%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright ©1999-2017 diyAudio