Very simple quasi complimentary MOSFET amplifier

RC,

Yes, I have used the DMM5401 but be aware that they are VERY TINY. At my age, I have a problem seeing the buggers..... I moved to singleton input stages (a la the JLH) some time back and so the thermal connection is not required. I was very pleased to give it away, they are, to use a description of our good friends the Yankees, 'pesky an' ornery' to fit.

I have just looked over your quasi circuit. I would foster your thoughts on this. Quasis are asymmetrical output stages, with two n devices in totem across the rails and they create a lot of H2 and H4. A lot of people actually love quasis.......

You have the emitter of Q4 on the output rail. You might put a 4004, a 22nF filmcap and a 100R resistor, all in parallel with each other, in between Q4 emitter and the output rail. This makes it very easy to set the bias current in the output stage and softens the switching transistors between the output devices under load.

Cheers,

Hugh
 
Last edited:
RC,

Yes, I have used the DMM5401 but be aware that they are VERY TINY. At my age, I have a problem seeing the buggers..... I moved to singleton input stages (a la the JLH) some time back and so the thermal connection is not required. I was very pleased to give it away, they are, to use a description of our good friends the Yankees, 'pesky an' ornery' to fit.

I have just looked over your quasi circuit. I would foster your thoughts on this. Quasis are asymmetrical output stages, with two n devices in totem across the rails and they create a lot of H2 and H4. A lot of people actually love quasis.......

You have the emitter of Q4 on the output rail. You might put a 4004, a 22nF filmcap and a 100R resistor, all in parallel with each other, in between Q4 emitter and the output rail. This makes it very easy to set the bias current in the output stage and softens the switching transistors between the output devices under load.

Cheers,

Hugh

Thanks for dropping in Hugh. I've seen this resistor in this configuration in Quasi's NMOS project and wondered about its function; however, his design doesn't have the diode or bypass cap.

I'll include these in the redraw I'm about to post shortly.
 
Latest redraw incorporating Mooly's revised R values for the tail resistor and feedback network, tail resistor and output gate stoppers, as well as Hugh's suggested changes.

I've also tidied up the front end, band-passing the signal (added 330p ceramic), adding a 100k resistor to prevent any big bang in the speaker if the input is accidentally connected while powered on, and adding inverse parallel diodes around the ground lift to pass fault current.
 

Attachments

  • CSH2.png
    CSH2.png
    14.5 KB · Views: 3,896
Didn't peter baxandall have some diode trick that improves linearity ? Maybe worth a google. Or maybe Hugh's post covers that already.

Edit : worth checking if the trick is for BJTs so maybe needs more diodes for a fet ops ? Blind speculation omp
 
Last edited:
Administrator
Joined 2007
Paid Member
Mooly, I've never used Spice, so I don't know how to interpret the result. Is it good or bad?

Erm :D I would say however it appears and whatever anyone says... its no real guide to how it will sound :)

The result shows a spray of harmonics, but the 2nd harmonic at 2kHz is dominant, and many believe that contributes to a musical sounding amp. It could well sound very good indeed.

If you want to install Spice and run the file I'll give you some pointers on how to do that... and then you can experiment with it.
 
Administrator
Joined 2007
Paid Member
No pressure :D

That's it. The slope on the FFT is caused by the larger caps, so to eliminate that effect you can make those caps absolutely massive such as 800,000uf each. Thats the input cap, the bootstrap and feedback caps. Also run the sim for longer such as 800ms rather than say 8ms. When you run the FFT choose to do the sample on the last part of the run like this.
 

Attachments

  • FFT Settings.JPG
    FFT Settings.JPG
    239.3 KB · Views: 1,929
I will likely follow in Gareth's footsteps and lay this out using a combination of SMT and TH devices.

I stumbled across this part in a SOT-32-6 outline and thought it would be ideal for the LTP since it comprises two "super matched" 5401 NPN transistors in a single package. A quick search here revealed that AKSA have used this in some of their designs.

http://www.diodes.com/datasheets/ds30437.pdf

Are there any downsides apart from the reduced Pd, which won't be an issue for me in this application. Frankly I'm surprised this device isn't more widely used considering how important a close match is for the diff amp transistors.


Hi,

I used these very parts, they do offer very nice performance and closely coupled complimentary parts. see here: http://www.diyaudio.com/forums/solid-state/196973-tgm5-all-bjt-simple-symmetric-amplifier.html

However, I used them on a pcb I made in the kitchen sink and it was quite a challenge to solder them without a proper silk screen etc. In the end I got somebody more expert than I to solder these tiny parts for me. I decided that I'd never be able to repair the boards if one of these tiny devices was damaged. So I decided never to use them again.

I had the same good advice from Hugh and he's right. It's best not to use them.

For mixed SMT and TH my findings are
a) do use SMT resistors and diodes,
b) do NOT use SMT parts with more than 3 legs on them which means these tiny matched parts with 6 legs are a bad idea,
c) there's very little to be gained using SMT transistors when the TO-92 parts don't take up any more space and being through hole you get a set of via's thrown in. This makes layout easier. I don't use SMT transistors as a result,
d) don't use SMT capacitors with legs that fold underneath as they are not suited to hand soldering, and be careful of SMT film caps as they can be overheated during soldering, I do use chip-capacitors, especially ceramic (which are quite good in NP0 type)



Keep going !
 

Attachments

  • completed set.jpg
    completed set.jpg
    278.4 KB · Views: 2,058
Last edited: