MOSFET gate voltage

[xplod1236]

I am designing an amp that will run off of +/-75v rails. I want to use MOSFETs for the outputs, but I don't know much about them. How much voltage do I need at the gates to get the FETs to swing from rail to rail (or close to the rails)?

 

[Mr Evil]

Depends on the configuration and the devices. If they are in common-source configuration then it's easy to get them to swing to the rails with Vgs of maybe 5-10V or so, limited only by the on resistance. If they're configured as source-follower then you will probably need to take the gate voltage a couple of volts above the rails to get maximum swing, also limited by on resistance.

 

[xplod1236]

What's the difference between common-source and source-follower?

 

[richie00boy]

Common source = CFP (complementary feedback pair / Sziklai pair), n-ch device in lower half and p-ch device in upper half.

Source follower, n-ch device in upper half and p-ch device in lower half.

 

[xplod1236]

Attached is the amp I am working on. I need some help chosing output FETs. Which ones would be good? Would it be better to use N-channel on both rails? If so, how would I go about modifying this circuit to use all N-ch output FETs?

 

[EchoWars]

That's a really small amp...microscopic. in fact.

 

[xplod1236]

Thanks for pointing that out. I haven't realized that I didn't attach it.

 

[AndrewT]

Hi,
I don't follow the workings of this schematic. It seems unconventional to me.
Comments anyone??
Try looking up previous threads here for suitable versions, there are a couple running just now in solid state.
regards Andrew T.

 

[richie00boy]

I've seen this kind of thing before, it's not common. Common base topology may be a way to describe it. It may be similar to Jan Dupont's (ACD) Lynx amp.

The output stage consisting of Q3,4 M1,2 is wrong as there's no method of biasing the output devices. R4 is doing nothing. If it makes it look OK in simulation that's because the output stage is incorrectly set up in the first place.

 

[Upupa Epops]

All this schematic is uncorrect, please read something about design of amps - I recomend to you for example pages of Erno Borberly .

 

[peranders]

I am designing an amp that will run off of +/-75v rails. I want to use MOSFETs for the outputs, but I don't know much about them. How much voltage do I need at the gates to get the FETs to swing from rail to rail (or close to the rails)?

I'm afraid that Upupa is right. You must get basic knowledge first because your efforts so far is totally wrong, sorry.

You can find some inspiration at Rod Elliot
http://sound.westhost.com/

Anthony Holton
http://www.aussieamplifiers.com/

http://www.borbelyaudio.com/
http://homepages.strath.ac.uk/~cnbp111/amp1.html
http://www.analog.com/UploadedFiles/Application_Notes/492001115525484056221917334AN211.pdf
http://mirand.dk/
http://users.ece.gatech.edu/~mleach/lowtim/
http://www.astro.uu.se/~marcus/private/m250.html
http://www.ne.jp/asahi/evo/amp/

And many many more.

 

[xplod1236]

That is a modified schematic from national. R4 is there to take take out crossover distortion. The opamp compensates for the outputs not being biased.

I ran this in a simulator, and it seems to be working fine.

 

[peranders]

You are forgetting about temp and bias stability issues. The parts you are using are indentical and at the same temp and this is not the case in real life.

Why don't you test your circuit in real life?

Do you have any reference from National where we can find more info?

 

[Mr Evil]

The circuit isn't completly wrong, it will work. Because the voltage swing required from the op-amp is very small, and the common base stage is very fast, crossover distortion can be quite low even with no bias. I have built similar output stages before and they work nicely. You have to be careful of stability though, since it has a high voltage gain.

As for your questions: It's better to stick with complementary pairs since the symmetry will tend to be better rather than fudging one half of the output stage. Personally, my favourite MOSFETs are those from Exicon. They're quite linear, very robust,not too expensive and helpfully have a low Vgs threshold.

 

[xplod1236]

Do you have any reference from National where we can find more info?

http://www.national.com/an/AN/AN-272.pdf page 4 and 5 (+/- 120v swing booster). I would've posted that this morning, but I did not have time to find the link.

You have to be careful of stability though, since it has a high voltage gain.

The gain is only 10x, so would I still have problems with it?

crossover distortion can be quite low even with no bias.

This is for a sub amp, so a little bit of crossover distortion that's not audiable would be fine.

 

[Mr Evil]

...The gain is only 10x, so would I still have problems with it?..

I mean the output stage has a high open-loop gain. That means that the overall gain could be greater than 1 when the phase shift exceeds 180 degrees, and the amp will oscillate.

The 100pF feedback around the op-amp will help with that, but whether or not it is enough may depend on what devices you choose for the output stage. Make sure to simulate it thoroughly.


I was wondering if you really need Q3/Q4 at all. You could probably get away with putting the MOSFETs where Q3/Q4 are now. That would make it much easier for you to get rail-to-rail output swing.

 

[Zero Cool]

You know, looking at your schematics, you could easily copy something like the Soundcraftsmen circuit design, or just about any commercially available amp for that matter. the parts count would be about the same and you have a proven design that works with the added benifit of the circuit being very clean!

I think your working way too hard to achieve the same goal.



Zero

 

[Zero Cool]

For example, check out the schematics found on the exicon website for there demo boards.



There is some great information here about the design as well.

http://www.profusionplc.com/static/images/data%20sheets/apnotes.pdf

Zero