Introducing the Simple Opamp / Mosfet Amplifier (SOMA)

Hi guys,

I've been sulking around on this forum for a while now but only recently signed up. What prompted me to do this was my recent designing and building of a simple amplifier which I was really impressed with and thought I'd share.

Very few (in fact more likely none) of the ideas in this amplifier are new but the overall effect is a good sounding amp at a very good price. My design criteria was simple, I wanted a direct coupled, active ground amp which could easily drive a pair of low impedance headphones to levels higher than I'd ever want to listen to. The result I've christened SOMA (for Simple Opamp/Mosfet Amplifier), following the much loved four letter acronym naming convention.

I hope that some of you see fit to build this amp for yourselves so you can let me know what you think.

Mike.

P.S. The power supply was built as two separate 9V supplies which are tied together in the amplifier chassis with the mid-point being 0V. This gives the V+ and V- rails (+/-9V). This was done to try and provide a level of modularity to the PSU.

Hmm, for some reason the images appear to have been resized, they were originally 800x600 GIFs. I'll try and fix them.

I wasn't aware of PRR's tori, but you're right, it is almost the same amp, the differences being trivial things like the bias levels, bypass caps and voltage gain, and the fact that mine was built with an active ground.

The circuit was built on vero (AKA stripboard or protoboard depending where you are) though with careful attention to layout. I run all power supply connections back to a common star point and have decoupling caps as close as possible to their respective components. The MOSFETs used are IRF620's.

I did originally design the circuit with a feedback compensation cap but on building the amp left it out initially and testing showed no oscillation on the output, even when driving a 2Vp-p 10kHz square wave into a 32ohm load. I must admit though that this amp was designed and built to be used with my Grado SR80i's hence the relatively low rail voltage. It also means that I've not tested the amp with other loads which may affect the stability.

I'm interested to know what benefits could be gained by using a MOSFET as the CC sink? My main reason for using the LM317 was purely due to its simplicity.

I did consider prebiasing the MOSFET by making R9/R11 1k and then adding a 1k trimpot pulling the gate up until the output voltage was 0V but decided against it for two reasons. 1) you still need to close the feedback loop in order to make sure the output stays at 0V and doesn't drift as the MOSFET and LM317 heat. 2) because the opamp is continuously driving the gate to about 4.5V (to maintain a 0V output) it is biased well into class A. My like of class A is purely a cosmetic thing, I love the idea of all that power (15W total including the PSU) just so I can drive a set of headphones with a few tens of mW (under normal listening conditions, though the amp can supply >150mW into 32ohm without clipping).

Thanks for all the feedback so far, I'm really interested to know what sort of gains you think I could make by replacing the LM317.

I'm aware of the loss of voltage but as stated in the above post (which I realise wasn't there when you posted yours) I designed this amp with my grado's in mind and they really don't need that much voltage.

As for the images, they were originally GIFs but when I uploaded them they were automatically resized and converted to JPG which really made a mess of things. In order to try and fix this I've converted the originals to JPG and reloaded them so they aren't stuffed with by the forum software. They look better now but I agree, they looked much nicer as GIFs.

Given the general opinion that the LM317 makes a poor CCS I've upgraded the CCS appropriately. I looked at the possibility of using a depletion mode MOSFET in saturation as suggested by dsavitsk but I found that MOSFETs with the desired IDSS are fairly rare and I really wanted the amp to be designed with easily obtainable parts. As a result I ended up with the zener referenced MOSFET based CCS shown.

As per nikongod's suggestion about adding a feedback compensation cap, I've done so also. 100pF sets the -3dB point at around 84kHz with the -1dB point being at 40kHz. I never had any trouble with high frequency oscillation (or any frequency for that matter) but this measure should ensure it stays that way.

While I was making changes I also increased the rail voltage to +/-15V, increased the MOSFET bias current to a frankly slightly ridiculous 400mA, and adjusted the NFB slightly. The overall gain of amplifier is still quite low at 9.5dB (Vout/Vin = 3) but it can be easily increased by altering one resistor per channel (R3L and R3R) if I ever upgrade to a higher impedance headphone and need the extra voltage.

As always I look forward to your comments/suggestions.


Mike.