Amplifier Role-call / Preferences ??
Moderators: piaptk, tragwag, Steve E., Aussie0zborn
Amplifier Role-call / Preferences ??
Hey all
For anyone who'd like to share...
What sort of amplifiers are being used for stereo cutting... any preferences of amp class ??
For anyone who'd like to share...
What sort of amplifiers are being used for stereo cutting... any preferences of amp class ??
Re: Amplifier Role-call / Preferences ??
I am running a bridged class AB thing built as a bipolar EF triple with a folded cascode input stage, basically done that way so that if/when I add feedback the amp will have negligible impact on phase margin anywhere that matters. It measures flat in frequency and phase out to 125kHz where there is a deliberate second order rolloff, and will put north of 12A into the coil at 20kHz until the voice coil overheat breaker trips.
I would note that these drivers are highly reactive and the SOA is a much bigger headache then you would expect, I am running 12 transistors in each leg (6 PNP, 6 NPN) for a total of 24 output devices per channel, but it is utterly reliable (And probably rather overkill).
It is tempting to think that a class H topology would save both a lot of heat and would make the SOA headache go away, but they bring there own pain when very high power at HF is a requirement.
One other hint, the output snubber resistor in most amps is NOT dimensioned to support massive amounts of HF boost, it can be worth uprating this if repurposing a PA or hifi amp, makes smoke less likely.
I stole the circuit breaker (And series RC to compensate driver inductance at high frequency) from the SAL74, and sort of reworked it a bit to take advantage of modern doings and to make it run a better model of voice coil temperature as a function of current and measured value, got to love how easy microprocessors make that shit.
I would note that these drivers are highly reactive and the SOA is a much bigger headache then you would expect, I am running 12 transistors in each leg (6 PNP, 6 NPN) for a total of 24 output devices per channel, but it is utterly reliable (And probably rather overkill).
It is tempting to think that a class H topology would save both a lot of heat and would make the SOA headache go away, but they bring there own pain when very high power at HF is a requirement.
One other hint, the output snubber resistor in most amps is NOT dimensioned to support massive amounts of HF boost, it can be worth uprating this if repurposing a PA or hifi amp, makes smoke less likely.
I stole the circuit breaker (And series RC to compensate driver inductance at high frequency) from the SAL74, and sort of reworked it a bit to take advantage of modern doings and to make it run a better model of voice coil temperature as a function of current and measured value, got to love how easy microprocessors make that shit.
Re: Amplifier Role-call / Preferences ??
That's cool. Are using mercury wetted relays?dmills wrote: ↑Fri Sep 10, 2021 6:01 amI stole the circuit breaker (And series RC to compensate driver inductance at high frequency) from the SAL74, and sort of reworked it a bit to take advantage of modern doings and to make it run a better model of voice coil temperature as a function of current and measured value, got to love how easy microprocessors make that shit.
Re: Amplifier Role-call / Preferences ??
Back to back switching mosfets with SiLabs isolated drivers my man, the future is now!
Mercury displacement contactors are really, really cool technology, but we got better ways to do this today for the sorts of voltages and currents we care about.
I have also become a huge fan of the little Vishay opto mos AC devices for signal switching, ~30pF is annoyingly capacitive in a high Z circuit, but I also do mostly low Z designs, so it is not that much of an issue and they don't need whetting current to be reliable (Which removes a HUGE headache), highly recommended for digitally controlled analogue.
I actually do have a load (Like ~60 or so) mercury wetted signal relays kicking about waiting for the right audio project (I have been contemplating a mastering transfer console using the things, but, big, heavy and needs to be right side up).
Mercury displacement contactors are really, really cool technology, but we got better ways to do this today for the sorts of voltages and currents we care about.
I have also become a huge fan of the little Vishay opto mos AC devices for signal switching, ~30pF is annoyingly capacitive in a high Z circuit, but I also do mostly low Z designs, so it is not that much of an issue and they don't need whetting current to be reliable (Which removes a HUGE headache), highly recommended for digitally controlled analogue.
I actually do have a load (Like ~60 or so) mercury wetted signal relays kicking about waiting for the right audio project (I have been contemplating a mastering transfer console using the things, but, big, heavy and needs to be right side up).
Re: Amplifier Role-call / Preferences ??
I'll take look at those. I haven't heard of those drivers. Do you have to play any tricks in with the control signal to deal with the latency of the microprocessor or whatever you are using?
Re: Amplifier Role-call / Preferences ??
Si5751 and friends, 35us turn off time beats any relay.
Of course actually getting them at the moment is a whole other question!
With the micro getting around its inner loop at audio relevant rates (a few tens of kHz) my approach was quite simple, use the bridge measurement to calculate a rough value for temperature and hence winding resistance (Note the bridge is heavily lowpassed so this will be a bit laggy), then use that resistance plus the measured current value to compute the energy input over one sample time, hence delta T over the last sample, leaky integrate delta T and add to measured value, trip if result gets too high.
The advantage of the small micro, apart from making a whole mess of annoying discrete logic go away is that it makes things like I^2 really easy, and also makes driving a multiplexed LED bar easier then it would otherwise be (An LED bar or two is much cheaper then a moving coil meter today). The same micro also handles the soft start relay sequencing for the main amp power transformer, and amplifier monitoring so it does earn its keep.
Of course actually getting them at the moment is a whole other question!
With the micro getting around its inner loop at audio relevant rates (a few tens of kHz) my approach was quite simple, use the bridge measurement to calculate a rough value for temperature and hence winding resistance (Note the bridge is heavily lowpassed so this will be a bit laggy), then use that resistance plus the measured current value to compute the energy input over one sample time, hence delta T over the last sample, leaky integrate delta T and add to measured value, trip if result gets too high.
The advantage of the small micro, apart from making a whole mess of annoying discrete logic go away is that it makes things like I^2 really easy, and also makes driving a multiplexed LED bar easier then it would otherwise be (An LED bar or two is much cheaper then a moving coil meter today). The same micro also handles the soft start relay sequencing for the main amp power transformer, and amplifier monitoring so it does earn its keep.