2017. január 13. 15:19 - Gazz

MOSFET Poweramp circuit

As I mentioned earlier I designed some power amplifier for supporting my guitarist activity. What lead me to this hobby originated in the eighties. I was teenager then and had no money so I started to built simple guitar effects. Later as my knowledge developed, I begun making my own design what usually sounded better than the factory made circuits. Soon my interest turned to power amplifiers. I used a very poor sounding Regent 50G what was available then for the little money I had. Furthermore my father asked me to repair an old radio what used germanium power transistors which died. So I had two goals.

1. building a better guitar amplifier

2. repairing my father's old radio.

First I tried traditional schematics what helped me in the case of the old radio but didn't improved my guitar sound. As I examined deeply the problem I found that the traditional designs almost always use the power transistors as emitter followers, see below.

 simple-100-w-power-amp.jpg

This is one of the most basic design of this area of electronics. In this configuration the power transistors amplify only current instead of voltage and current. Even it's true if complementer darlingtons are used instead of simple power transistors.This quite differs from the analogue case of tube amplifiers where the power tubes are allways used in grounded cathode configuration what results both voltage and current amplification.

Another problem with emitter follower configuration is that the power transistors need higher voltage drive than their emitters potential. It means that we cannot drive this kind of amplifiers's output to the whole range of power supply voltage without tricks. Necessarily its output must be less than supply voltage. If we applicate bootstrap then one power transistor's emitter ( T4) can be driven over the power supply voltage due to the fact that the bootstrap increase the driver voltage of this transistor over the positive supply voltage. But bootstrapping results some problem about the phase transmission and I hear it as a strange squelch in the sound.

Nowadays everybody uses power mosfet transistors instead of bipolars. The widespread configuration of their using is quite similar to what we see above with some minor modification. They are more reliable than bipolar power transistors and sound better. In the other hand power mosfets need way higher drive voltage between their source and gate pole than bipolar transistors between their emitter and base. So the headrom is less if we use them in the traditional configuration.

In my case I wanted to upgrade an old, transistor powered Marshall amplifier. It had symmetrical power supply voltage +20V and -20V. Well it is quite low so I need all volts for my poweramp's output to produce as much power as it can be possible. I decided to use power mosfet transistors instead of bipolars due to their characteristic is similar to vacuum tubes. So I had to find a special configuration where I can drive my amp's output from negative supply voltage to positive supply voltage, so in the whole range.

It was obvious for me from the beginning that the traditional configuration did not provide this. After analyzing the circuit I realized that the way of using the power transistors must be changed. I need configure them in grounded source mode instead of source follower. In this case these transistors would amplify voltage too and so they need way less drive voltage for getting the headroom reaches the whole range between negative and positive supply voltage. This drive voltage can be provided easily. Another point that they must be stabilized thermally. The typical IRF type power mosfet's drain current is increasing by the temperature if the gate voltage remains the same. Studying the graphs I found that I must decrease the gate voltage by 5,5 mV per Celsius if I want the drain current being the same. Silicone transistors base-emitter opening voltage changes by -2mV so I have two ways. I can use three of them before every power mosfet or I can use one for only the one of the mosfets and I set the DC voltage circuit gain to six times. The negative feedback then arranges to share this compensation between the two mosfets. Of course I need more AC gain in the circuit but it can be arranged easily by condensators in a proper positions.

Another point was that I needed the default output level to 0V what is exactly the center between the positive and negative supply voltage. For this the most reliable solution is a differential amplifier (actually a long-tailed pair) at the front of the circuit. There are another options but none of them is stable enough. There is a debate about using long-tailed pairs in poweramps as it can be the source of annoying transient intermodulation distortion called TIM. In the other hand in the case of guitar amplifier which has a way narrower frequency range than a HiFi circuit this point is not relevant.

Though my design starts with long-tailed pair just like the traditional one the similarity ends right here. I use totally different thermal  stabilization and bias than the classical. Also I drive the mosfets different way which are in grounded source configuration contrary to the classical solution. This arrangement provides full power headroom, high thermal stability with simple construction.

Let's see the schematic then.

powermosdiff.gif

 As the Figure 1 can be seen as the basic schematic of bipolar amps I think this Figure 2 has the right to be the basic schematic of mosfet poweramp technology due to its simplicity and because it uses all of the advantages of mosfet power transistors.

Let's see how it works. Practiced eye instantly recognises the usual long tailed couple (T1-T2) at the front of the schematic which works here as phase splitter. We use a current generator as emitter feeder for two goals. One point is it increases the amplifier's stability and symmetrical working. Second point that - instead of the practice of common bipolar amp circuits - we use it for biasing the output stage. My theory is that poweramp's DC gain shouldn't be enormous because we don't use it for amplifying DC but AC. Even more important principle that the power mosfets must be those parts where the biggest voltage (and current) amplifying should happen. For this goal I kept the DC gain of earlier stages as low as about 12 times. The long tailed pair amplifies only twofold and the second stage (T4,T8) do only less than sevenfold. Actually this is very good for thermal compensation. T10 is a transistor what must be applicated onto the heatsink the power mosfets are bolt on. So it is connected them thermally. Every celsius grade increasing decreases the voltage on this part by 2 milliVolt. To stabilize a power mosfet thermally it needs 6 millivolt decrease on its gate by every celsius grade what means we have to amplify the decreasing of T10 voltage threefold. In the other hand it is enough only for the one of the two power mosfets (T5) so actually we need sixfold DC amplifying after T10 then the negative feedback from the output takes care to share this compensation between the two.

As we see we drive the power mosfets by emitter followers (T6,T11). These provide low impedance which is good for fast charging and decharging the input capacitance of power mosfets (about 1-5 nF). In the other hand we use these capacitances to set the upper limit of frequency range this amplifier is working with. The human ear does not hear frequencies higher than 20kHz. It does not make sense to let this amplifier work with frequencies we even don't hear. Furthemore amps tend to produce unwanted feedback on higher frequencies what cases bad and unpredictable behaving of the circuit. So it is strongly suggested to limit the circuit to work only with that frequency range we hear. Or at least not so much wider range... That is why we drive the mosfets through R9 and R6 which consist low pass filters with the gate capacitances of the mosfets.

We have not discussed only one interesting part of this circuit yet. This is a special arrangement of T8 and T9 transistors. It is quite rare to find this in circuits though it is not particularly complicated. 

T8 seems to work in emitter follower mode. Actually it is wrong. It works in a grounded emitter mode but in this special case the ground is a floating one. The matter is that T1's collector sees T9's emitter as ground and the voltage what is a result of T1's current flows through R2 is relative to T9's emitter what is driven by that voltage what T8's current makes on R8 what depends on how much voltage of R2 drives T8 relatively to the voltage of R8.

Actually it is a kind of bootstrap but it uses transistor (T9) instead of capacitor. It results that T8 amplifies the voltage of R9 on R8 but contrary to the classical grounded emitter configuration it does it in same phase. The DC gain is the ratio of R8 and R1, the AC gain due to the C3 capacitor is close to 500, depends on the beta of the transistor.

We use this solution because we wanted simmetry between the two driver branch what lead to the two power mosfet. T4's role is unavoidable in this construction so I had to find a similar construction but with same phase amplifying in the other branch. That is why I used this special bootsrap solution.

The circuit is stable and relatively simple. Its sound differs to usual mosfet poweramps but is quite close to the famous VF2 circuit though that is not so detailed.

Let me know if anybody has questions. And sorry about my bad english.

 

 

 

 

 

Szólj hozzá!

A bejegyzés trackback címe:

https://gazz.blog.hu/api/trackback/id/tr1312083287

Kommentek:

A hozzászólások a vonatkozó jogszabályok  értelmében felhasználói tartalomnak minősülnek, értük a szolgáltatás technikai  üzemeltetője semmilyen felelősséget nem vállal, azokat nem ellenőrzi. Kifogás esetén forduljon a blog szerkesztőjéhez. Részletek a  Felhasználási feltételekben és az adatvédelmi tájékoztatóban.

Nincsenek hozzászólások.