Tír Na HiFi

This was hived off from the "mosfets with a twist" thread

There is another very interesting approach using mosfets as followers as current gain devices - Susan Parker's Zeus amplifier
It's not SE but push pull & uses an input transformer as voltage gain device & output transformer (along with MOSFETs) as current gain stage. It's also a high PSRR design & said to sound wonderful but this should be taken to another thread if we want to discuss?

nige2000 wrote: ↑Mon Jul 10, 2017 9:01 pm think if i were to use transformers id like to use them with a good step down probably 5:1 or greater
maybe it was the ones i was using but at anything under 2:1 in comparison to 5:1 and greater seemed to be a loss noticeable in tautness in bass notes and in resolution

so in which case id need to add gain to input signal say x10 and use a 5:1 transformer at the front of that zeus amp,
id be interested in that, id guess the transformers wont be cheap

From her website I just noticed she mentions an SE version of the amp

Variations
I have bread boarded a single ended version of this design, reconfiguring T1 for a single secondary to drive one power semiconductor (Q1) and connecting both T2 primaries in series . The principle stays the same but the output transformer must be capable of taking the DC current without saturating, and the voltage reference to T1 secondary must be better regulated as any noise here will be coupled to the output by transformer T2 rather than cancelled out by it. It works at a quarter of the power of the push-pull version, but I didn't make any measurements as to distortion, etc.

As you mentioned, Nige, the input transformer ratio is a stepup 1+1;10+10 & expensive

8160 Zeus Power Amplifier by Susan Parker Input Transformer

1+1:10+10 Fully geometrically balanced input transformer. Designed for this excellent fully balanced MOSFET design with no overall negative feedback. Size A. U-Clamp mounting.

Total Price:  £115.74

Output transformer:

9840 Zeus Power Amplifier by Susan Parker Output Transformer

4 + 4 Quad-Filar windings Designed for this excellent fully balanced MOSFET design with no overall negative feedback. Size N

Total Price:  £220.58

But she has specified using some PS toroid transformers for input & output for proof of concept (& apparently pretty good sound) - I'll look for the specs of these recommended PS trafos

Sounds like hassle

nige2000 wrote: ↑Mon Jul 10, 2017 11:52 pm Sounds like hassle

Really? - after It Hasn't Exploded Yet, amplifier build?

Off the shelf toroids can be used although not as low distortion as the Sowter ones
You probably have a shed load of these in your stores, Nige? :)

Input Transformer - 30VA
Bandwidth: 10Hz to 24 kHz (-3 dB), 28 kHz (-6dB)
Spec: 30 VA, Pri single 240 : Sec 15 + 15 volts at 1 amp
Inductance of  240 V winding = 9.19 H @ 1 kHz, 26.3 H @ 120 Hz.
Impedance of 240 V winding = 339 KΩ @ 1 kHz, 112 KΩ @ 120 Hz.
Inductance of each 15 V winding = 31.5 mH @ 1 kHz, 273 mH @ 120 Hz.
Impedance of each 15 V winding = 1.2 KΩ @ 1 kHz, 620 Ω @ 120 Hz.
Rterm: 1.8K - 100R trim to Vbias - 1.8K (total 3.7K).
Driving impedance = 80 ohms Series, 20 ohms Parallel.

OR

Input Transformer - 15VA
Bandwidth: 10Hz to 25.5 kHz (-3 dB), 31.5 kHz (-6dB)
Spec: 15 VA, Pri 115 + 115 volts : Sec 12 + 12 volts at 650 mA
Inductance of  115 V winding = 4.6 H @ 1 kHz, 20.9 H @ 120 Hz.
Inductance of each 12 V winding = 96.2 mH @ 1 kHz, 586 mH @ 120 Hz.
Impedance of each 12 V winding = 1.3 KΩ @ 1 kHz, 903 Ω @ 120 Hz.
Rterm: 3.6K - 115 + 115 centre tap to Vbias.
Driving impedance = 59 ohms Series, 15 ohms Parallel.

Output Transformer
Spec: 217 VA, Pri 120 + 120 (not used) : Sec 31 + 31 volts at 3.5 amp
Inductance of each 31 V winding = 95 mH @ 1 kHz, 541 mH @ 120 Hz.
Impedance of each 31 V winding = 1.77 KΩ @ 1 kHz, 922 Ω @ 120 Hz.
Output impedance in circuit when powered = 4.3 Ω @ 1 kHz.

15VA size for the input transformer is huge! I reckon a ferrite cored one on PQ32 would be quite sufficient.

The primary weakness of the Zeus amp is it really needs a stonking good buffer to drive its input trafo. The input inductance (for 15VA) of about half a henry is quite demanding of LF current. But also the capacitance on the secondary is multiplied by over a 100X due to the step up ratio. So this means the buffer's heavily loaded at HF too. Far better to have a custom signal (not power) transformer here.

The output transformer as specified is way over-sized. I'd suggest no bigger than a 50VA toroid.

abraxalito wrote: ↑Tue Jul 11, 2017 12:52 am 15VA size for the input transformer is huge! I reckon a ferrite cored one on PQ32 would be quite sufficient.

The primary weakness of the Zeus amp is it really needs a stonking good buffer to drive its input trafo. The input inductance (for 15VA) of about half a henry is quite demanding of LF current. But also the capacitance on the secondary is multiplied by over a 100X due to the step up ratio. So this means the buffer's heavily loaded at HF too. Far better to have a custom signal (not power) transformer here.

The output transformer as specified is way over-sized. I'd suggest no bigger than a 50VA toroid.

Right Is this why the figure of merit in choosing the mosfets for this amp seems to the one with lowest Crss?
They settled on STW34NB20 - "Crss: Reverse Transfer Capacitance = 90 pF"

What you say seems very reasonable

jkeny wrote: ↑Tue Jul 11, 2017 1:44 am Right Is this why the figure of merit in choosing the mosfets for this amp seems to the one with lowest Crss?
They settled on STW34NB20 - "Crss: Reverse Transfer Capacitance = 90 pF"

The Crss is most likely the most relevant capacitance in a source follower configuration though it does depend on the application.

I note that the gm isn't too impressive, about 2S @ 1A. This bootstraps the input capacitance (2400pF) to some degree, but let's look at how much. The output impedance will be 0.5ohm (assuming 1A bias) hence the bootstrap effect will be the ratio of the Zout to the Zload. Given we're running balanced the Zload is 4ohm. We can therefore divide 2400pF by 8 to get the effective input capacitance - 300pF. So in this application the input capacitance is not negligible, it does exceed the Crss due to the poorish gm. But the input transformer's self-capacitance is going to matter too, it'll be in the region of 10's to maybe 100pF. Probably not higher than the 300pF input capacitance so we can safely say the input capacitance dominates.

What kills the driving source is the effect of the 10:1 step up ratio - this makes the input capacitance look like 30nF. That much capacitance looks like ~270ohm @ 20kHz.

We can do a quick comparison with the TK13E25D - I am first of all making an assumption that this FET can handle the power dissipation, its only a TO220 so would need excellent heatsinking. Assuming that it can then its input capacitance is less than half that of the STW (1100pF vs 2400pF) and its gm is about double (4S vs 2S). The Crss is about 12pF (vs 90pF) but somewhat more non-linear. Overall looks like a better choice provided the output power is manageably low.

Here's a prototype speaker amp coming together based on the 'hot buffer' concept. Four channels, I'll start by arranging them as two pairs of two for about 7W per channel, then when I have another case convert the first 7+7W amp to a 14W monoblock. The experiment is to test how well the amp case can cool itself, the limit on the power output is going to be how hot the case gets. The 7+7W is based on an estimated 56W total (25% efficiency).

Ah super cool

Funny how the requirement for heatsink brings out the case

What size are the fet packages ?

Hope it works out so I can get stuck in :)

Having trawled around Taobao a little, I came across this case which is 120mm tall (the one I'm building with now is 80mm) so should be able to shift a little more heat.... It also lacks provision for a volume control which is just perfect for a poweramp. The FETs are TO220.

Here's the seller - https://item.taobao.com/item.htm?spm=a1 ... 2142218594

Of all the amps I've built, this one was far and away the easiest to throw together and I think probably the only one that worked first time. ClassA amps with no GNFB are not liable to oscillations (one pitfall for inexperienced amp builders) and also they're relatively intolerant of wiring topology to get excellent sound. So I reckon this amp is a very good beginner's amp for anyone who's not gotten their togs wet in amp construction. Its only a buffer right now, needs a voltage gain stage (about 20dB) which I'm working on. My current design for this gain stage is considerably more complex than the buffer, I'll experiment with a 10:1 step up transformer in the spirit of Zeus.
Just a lash-up at the moment, running from my bench supply (2A, 24V). Eventually I will fit a switching brick inside and the voltage gain module. Even with those added it'll still be mostly air.