On amps with the twin pots, bias/offset, will it work to have two bias 'power' settings that get routed through a rotary selector switch? I'm thinking about doing this on an F6. The idea being to have the option to run it on low power in terms of total power consumption. Or is it not possible on an amp like this to operate at much lower ACA Mini levels of bias/total power?
I'm not so sure it would work tbh. If you alter the bias (which is one preset) then the output voltage will shift as well. Both preset values would need to shift together. So still doable 🙂
Why not set it to a lower bias and reset the output to zero volts and see if it meets your needs for power output and audio quality. A quick test first like that is well worth doing.
Any switch would be best wired (imo) to shunt a new value resistor across what is already there. In other words set the normal high bias and zero offset and then determine a new value to shunt across the wiper and bottom end of both presets to give your new bias and zero offset. Use a 4 pole switch for controlling both channels.
Why not set it to a lower bias and reset the output to zero volts and see if it meets your needs for power output and audio quality. A quick test first like that is well worth doing.
Any switch would be best wired (imo) to shunt a new value resistor across what is already there. In other words set the normal high bias and zero offset and then determine a new value to shunt across the wiper and bottom end of both presets to give your new bias and zero offset. Use a 4 pole switch for controlling both channels.
Yes, that is what I was thinking too -- I realize there would need to be tailored values on each end, bias/offset. We are in agreement on how to go about configuring a switch. The thing about experimenting on the amp is I haven't built it yet. I'm ready to make a Mouser purchase for the components and am debating this mod and if I should buy the switches rather than pay double on shipping after I build the amp and decide it will work.
If the amp will output 5w in class A at some 'biggish' reduction in power, say half, that is definitely worth implementing. The inspiration came from the MINI and I was considering building one for it's low power consumption (thinking batteries here) but it occurs to me, why not just run the big brother at low power?
I think you will have to try it. It looks to me like the output stage won't behave in the same way as say an ACA might at low bias. This one looks like it will revert to something more like Class B with an increase in odd order harmonics as you try and get more level from it in an under biased state.
You need to experiment with it. A lot depends what sort of levels you ultimately want to listen at.
You need to experiment with it. A lot depends what sort of levels you ultimately want to listen at.
Last edited:
Thank you, Mooly. That is the sort of thing I wondered about but I don't have the knowhow to analyze the system. I can say from experience that an M2x will operate down to quite low voltages and sound good. The unique bias circuit may be compensating for the lower voltage however, so I'm not sure if there is an actual decrease in total power consumption with that particular amp. Like you say, I'll just have to try experimenting with the F6 and see. If it becomes unhappy then I suppose the next best bet would be to just wire the Mini to the selector and completely switch amps 😀
I sort of got mildly interested in this 😀 as much from simulating with its coupling transformers as much as anything else.
Please remember this is just a simulation and the interstage coupling transformer will not be accurately modelled. I've added the .asc if anyone wants a play. All default models, should click and run.
Arbitrary levels set for bias and output voltage. In order from 2A, 1.6A, 0.4A and 0.16A bias currents. Final two images are distortion @ 2A and 0.16A bias.
Edit... pictures haven't attached in the right order. Showing as 1.6A, 2A 0.16A and then 0.4A
Distortion at 2A Bias:
Distortion at 0.16A Bias:
Please remember this is just a simulation and the interstage coupling transformer will not be accurately modelled. I've added the .asc if anyone wants a play. All default models, should click and run.
Arbitrary levels set for bias and output voltage. In order from 2A, 1.6A, 0.4A and 0.16A bias currents. Final two images are distortion @ 2A and 0.16A bias.
Edit... pictures haven't attached in the right order. Showing as 1.6A, 2A 0.16A and then 0.4A
Distortion at 2A Bias:
Distortion at 0.16A Bias:
I'm pleased the LT stuff was useful. The little interstage transformer is a bit of an unknown. Apparently it has 28 ohm DC resistance but I could not find any inductance listed for the windings and so I've guessed. Currently set at 60mH which gives a reasonable low frequency response.
Ok, I did some refreshers on your tutorials and was able to more or less make tweaks to the circuit. Very cool stuff! I tried to get closer to the actual amp spec but it didn't want to play along 🤔
I noticed you went with an alternative MOSFET to the IRFP240. When I tried to use the 240 it was basically incoherent. I'll keep tinkering and see if I can figure something out. I know the genuine F6 is rated for .05% THD at 1W, and that is accomplished at 1.55A bias, via 1.2V. Granted, this is from the SemiSouth R100.
How are you determining what the actual bias current is, are you measuring it?
I noticed you went with an alternative MOSFET to the IRFP240. When I tried to use the 240 it was basically incoherent. I'll keep tinkering and see if I can figure something out. I know the genuine F6 is rated for .05% THD at 1W, and that is accomplished at 1.55A bias, via 1.2V. Granted, this is from the SemiSouth R100.
How are you determining what the actual bias current is, are you measuring it?
Like this you mean. You need to increase the two gate voltage values to bring the current back to where you want:
Ha, yes! I was actually going the wrong direction. If you drop the voltage to 1.2 (according to authentic F6) it actually smooths out again, but the amplitude goes to millivolts. I didn't think to raise it the other direction. So at 5.4V, how do you determine what bias that translates to?
Edit: was that a hint using the other FET? That thing kinda rocks lol!
Edit: was that a hint using the other FET? That thing kinda rocks lol!
Last edited:
It's very interesting to see this stuff put into practice. I was struggling with the voltage disparity as it compared to NP's lecture on the F6, where he goes into detail on each parameter. After seeing how you arrived at the settings for the sim I retraced the steps for the SemiSouth 100R and sure enough, max gate threshold is 1.2V! Quite a satisfying exercise. To circle around to the original intent, the bias switch, my takeaway is that will be worth further trials on a physical amp. But-- the huge info boost you helped me gain is the real gem with this little side project 
You're on a roll, Mooly! I'm curious, what's your takeaway/interpretation? Also, what's your takeaway on attempting a switch, having tackled the sim'd version?
I would say the switched version is definitely worth a try if you want to make a cool running version. You should try listening first to some arbitrarily set bias points and see how you think it performs sonically.
As it stands you have around 35 watts dissipation per transistor at 1.6A bias. Reduce that to say 0.8A that falls to just 18 watts per transistor, a massive difference. Given that most listening is only at a watt or two it is worth having a play imo.
The sim would be good to set up with some switches to simulate what you want to do. I haven't time just now to do that but it should be doable. Replace the voltage sources with the Zener and put a couple of switches in to add your feature.
🙂
As it stands you have around 35 watts dissipation per transistor at 1.6A bias. Reduce that to say 0.8A that falls to just 18 watts per transistor, a massive difference. Given that most listening is only at a watt or two it is worth having a play imo.
I've never heard one so hard to say how it translates from what the sim shows to how it sounds (and how it really measures for real). The sim is great for showing trends though if not absolutes.
The sim would be good to set up with some switches to simulate what you want to do. I haven't time just now to do that but it should be doable. Replace the voltage sources with the Zener and put a couple of switches in to add your feature.
🙂
Oh good, I'm glad my thinking is not too far off base. What I ended up doing parts wise is getting an assortment of really nice NOS precision trimpots, big 5w wirewound beasts. This will give me easy adjustment access to bias/offset voltage and even one for feedback. And finally a couple of three position switches. The idea is to have maximum flexibility so I can readily tinker and learn what works and what doesn't. Plus I want it to be more of a platform for amps of various circuits rather than purely one single topology. Check out these babies--
I went with 500R for the feedback adjuster and I got 10k for the bias/trim for ten dollars a pop!
I went with 500R for the feedback adjuster and I got 10k for the bias/trim for ten dollars a pop!
Wow, they look quite the business 👍
On a practical level there are a few considerations to bear in mind. Any switching of bias would probably have to be done with the speakers disconnected simple because the switching would not be instantly simultaneous for both pots. That would create a voltage transient across the speaker. It would be more a switch for occasional use, perhaps to suit ambient conditions as much as anything. As Doug Self once said, a 'summer/winter' bias mode.
On a practical level there are a few considerations to bear in mind. Any switching of bias would probably have to be done with the speakers disconnected simple because the switching would not be instantly simultaneous for both pots. That would create a voltage transient across the speaker. It would be more a switch for occasional use, perhaps to suit ambient conditions as much as anything. As Doug Self once said, a 'summer/winter' bias mode.
You don't want anything that will pick up noise. The feedback network in this amp is all very low impedance and so should not be an issue but it is something to keep in mind.