Any interest in new Class AB power amp design/build with MOSFET outputs?

[xandresen]

Teaser alert!

I have extended the VHex+ power amp design of Valery Zaichenko into the VHex_step-up design, 50-70W into 8 ohms.
Step-up features lower distortion, current limiting, output muting, etc., while still having a relatively low parts count.
Have prototype PCB boards (one PCB per channel).

I know most amp builds are posted in DIY Audio's Solid State forum. However it has over 2000 pages of posts!! Using its search, I can't even find amp designs I know are there unless I search the name of the design.

So my query is:
Would HTG viewers be interested in following a build/evolution of a classic AB power amp design with vertical MOSFET output devices?

One evolution could add another pair of output devices for 100W into 8 ohms, 150W into 4 ohms.

And do any of you know how to solder small parts onto PCB's and read a schematic? because that's required for a build.

Thanks for your feedback!

PS: The design is Influenced by Bob Cordell's 700 page "Designing Audio Power Amplifiers", my experience as an Analog Electrical Engineer, and plenty of Spice simulation.

[DaveFred]

I have always wanted to build a powerful (>100 watt at 8 ohm) lateral mosfet AB amp, that is, if Exicon still has anything left...

[Reet]

I encourage you to continue your adventured in amplifier design, and look forward to reading about your progress. However, I don't think you will find many EE here to provide much valuable insight other than encouragement.

Myself, I'm an Electronics technologist, but don't spend much time digesting analog transistor circuits other than troubleshooting for repair and optimizing performance through simple / basic modifications. I'm of the mindset that others far more educated than myself on circuit design have already designed well functioning amplifiers that can be purchased cheaply as drop-in modules, so just use what is already available, I personally don't see a need to re-invent the class-AB wheel in 2025.

For this reason, my recent DIY amp builds were done by cobbling together parts from AliExpress, and pleasantly surprised with the resulting performance obtained for rather low investment and a fun assembly project. My amps were build on LJM modules, which are a BJT design based on Doug Self's blameless topology. Amazing performance at a low cost I must say! You can read about it here:

https://www.htguide.com/forum/forum/mission-possible-diy/959863-diy-amplifiers-ljm-modules

I'm now in the process of repairing some Adcom amplifier boards, which honestly look to be a similar circuit design, and perform similarly too. I am surprised with the Adcom amplifier, it includes some sort of protection circuit, but there is no protection relay at the speaker output, quite dangerous to your speakers if one of these things fails, you may end up with 50VDC at the speaker.

Look forward to your design.

[xandresen]

I have always wanted to build a powerful (>100 watt at 8 ohm) lateral mosfet AB amp, that is, if Exicon still has anything left...

Lateral MOSFETs need a different bias spreader and output stage than the vertical MOSFETs I study.

I believe DIY Audio has at least one lateral design. and their posters like high power...

[xandresen]

Hi Reet,

I did see your post on the amp modules.
In the spirit of audio cynicism, I suggest you look up Self's power rating for the circuit the LJM module is based on. Sometimes low cost products greatly stretch the specs.

As to "why re-invent the wheel" in 2025, think of the endless two and three way speaker designs that continue to pop up.

Personally I like to design electronics and always wanted to see if I could do a good power amplifier design. Compared to class AB, class D amplifiers are even more tricky for low distortion and require a matching power supply design to guarantee the amp will meet its specs (think Hypex). So class AB it is.

Speaker relays
seem in Cordell's book to be a double edged sword. The detector of large DC output voltage has to not trigger on full power 20Hz sine wave, meaning there is a delay before activating(releasing) the relay. And large power relays can take >10msec to release, depending on the relay and its driver design.
I end up not being certain the relay will protect the woofer. (not an expert here)
For the tweeter, my thought is it will be destroyed if an output device shorts to the rail - the tweeter's voice coil slamming into the hard parts in < 1msec, long before any relay can release. The crossover cap in series with the tweeter passes this high frequency spike.

Thanks for your thoughts.

[Reet]

What power rating would you give an amp with 4 pairs of 2SB817 / 2SD1047 output transistors, fed from +/-50VDC? For my use, I don't blast 250W sine waves through my amplifiers for power transmission, they are for Hi-Fi music playback. Given that +/-50VDC voltage rails allow for about 35VAC before clipping, and accounting ~12dB crest factor for audio, that leaves me with just under 10VAC on average, or 25W into 4 ohms. Do you think the 4 pairs of transistors can handle that, with brief bursts of 35V?

If you want to complete any more detailed analysis, the schematic of LJM L20.5 is here:



Endless speaker designs, yes, they are the part of the audio chain with the sound "character", endless possibilities and as a furniture element, much of a visual appeal element as well. But you're right, that wheel has also already been designed 50 years ago, and fine tuned over the years, and today the entire Hi-Fi market is mostly snake oil, re-hashing the same old designs with the latest high tech materials of the era.

Speaker relays - I'm quite confident that a woofer can take the 10ms pop, and I've no doubt the tweeter will survive too, well a normal dome anyway, so the relay addition for a DIY project is a "better than nothing" addition for not very much added cost. For normal voicecoils, there is thermal mass there that will take more than a few milliseconds to burn out. In my younger years I took some old 1970's radio speaker, I think 0.75" voice coil I'm sure it was rated for only 20W or so and plugged it straight into 120V wall socket For a "music video" sort of class project, cranking it to 11 style. It took a good 10 seconds for the speaker to burn out. The main problem with relays is that they're not very good at breaking large DC voltage, which is exactly what they'll be asked to do in case of disaster. People often under-size them. In my Aliexpress junk I chose 16A relays for 50VDC rails.

[xandresen]

Schematic thoughts
The output transistors are rated at 140V VCEO (this is an absolute max rating). The schematic shows 65V power supplies so the output transistors could see almost 130V. However we need to allow for maybe 130% surge in line voltage (if the supplies are unregulated). Likely some shorted devices at full power.
One could argue that the VCBO=160V is the more relevant rating here, but it requires the base to be hard grounded - the design is somewhere in between VCEO and VCEO.

However with +-50V supplies we have a voltage safety margin for the hypothetical 30% surge.

Just noticing there is only one buffer transistor stage between the voltage amplification stage (bias spreader) and the 4 output devices. This works but limits the open loop voltage gain. Not suggesting you change it - it's a major change. I wonder if Pass's design is for 2 output pairs?

Power expectations here seem strange to me. Four output pairs seem OK but isn't the heatsink a bit small, especially for the 4ohm load? Hard to tell from the photo.
approx 150W into 8 ohms, higher than 200W into 4 ohms.
Of course you have the luxury of being realistic and choosing based on playing music (with its crest factor). Amplifier designers are condemned to pass sine wave testing.
For 8 ohm load, heatsink gets hottest with about a 1/3 full power into load.
(side-note: pop music does not have much crest factor - it is adjusted to a nearly level volume.)

Amp stability
The amp has an output Zobel network. good.
You might consider a parallel resistor/inductor in series with the output to the speaker, to decouple the amp from any external high-capacitance load (like long lengths of shielded speaker cables, a terrible idea.).
You can add one if you want. Use an 2uH air wound inductor and a 2 ohm metal film resistor. values can vary 50%.

speaker protection
The tweeter concern I have is its voice-coil + dome rocketing into the mechanical stops. as in big crunch, not smoke.

agree on undersized speaker relays.
Cordell uses an RC snubber across the contacts to suppress most of the arc, even on larger relays.
He suggests 10uF NP electrolytic cap and 2 ohm resistor in series across the contacts. However he notes this will allow a small amount of audio leakage if the relay is also used to mute the amplifier.

[Reet]

Schematic thoughts
The output transistors are rated at 140V VCEO (this is an absolute max rating). The schematic shows 65V power supplies so the output transistors could see almost 130V. However we need to allow for maybe 130% surge in line voltage (if the supplies are unregulated). Likely some shorted devices at full power.
One could argue that the VCBO=160V is the more relevant rating here, but it requires the base to be hard grounded - the design is somewhere in between VCEO and VCEO.

However with +-50V supplies we have a voltage safety margin for the hypothetical 30% surge.

Just noticing there is only one buffer transistor stage between the voltage amplification stage (bias spreader) and the 4 output devices. This works but limits the open loop voltage gain. Not suggesting you change it - it's a major change. I wonder if Pass's design is for 2 output pairs?

Let's be clear, it is not Pass's design, or Doug Self's, but LJM's circuit.

+/-65V is "Absolute maximum", LJM does not recommend it, rather he suggests 40-45V, or "just as much as you really need" as any more is just wasted by increasing thermal demand.

Power expectations here seem strange to me. Four output pairs seem OK but isn't the heatsink a bit small, especially for the 4ohm load? Hard to tell from the photo.
approx 150W into 8 ohms, higher than 200W into 4 ohms.
Of course you have the luxury of being realistic and choosing based on playing music (with its crest factor). Amplifier designers are condemned to pass sine wave testing.
For 8 ohm load, heatsink gets hottest with about a 1/3 full power into load.
(side-note: pop music does not have much crest factor - it is adjusted to a nearly level volume.)

Heatsink too small? You can't be serious. Each heatsink is 300mm x 110mm x 45mm, with 7mm of solid aluminum before the fins, and directly bolted to the rest of the solid aluminum chassis. There's tonnes of thermal mass and heat dissipation available.

You must be thinking of TV commercials. None of the music I listen to have "level volume", most is at least 12dB CF. Pink noise has crest factor of 14dB, and M-Noise commonly used as a accurate test signal representing real world audio conditions, has a crest factor of 18dB. With that in mind, it should encourage anyone interested in hi-fi to get a big amp, not for sheer power but just for peak voltage output. You'd be surprised how much smaller amps are clipping under normal use. I mentioned before, I don't need 250W of output, I need 25W of output with short 250W peaks. 1/3 duty cycle is still quite generous, but definitely more realistic as far as "real world worst case".

Amp stability
The amp has an output Zobel network. good.
You might consider a parallel resistor/inductor in series with the output to the speaker, to decouple the amp from any external high-capacitance load (like long lengths of shielded speaker cables, a terrible idea.).
You can add one if you want. Use an 2uH air wound inductor and a 2 ohm metal film resistor. values can vary 50%.

I have one, it's 3.3uH || 10R.

[xandresen]

Guess I didn't see the heatsinks, just looked at the individual modules in your module post.

Peak power
3 way speakers i designed use a MiniDSP Flex8 digital crossover, so I need a separate amp channel for each driver.
Use 130W per woofer so as to not clip on classical music peaks, 50W for mid, 50W for tweeter. Similar power to your plan.