Power MOSFET at Audio?

[AB5NI]

Just wondering if any of u guys have ever played around with power MOSFET's at audio. In the ham-radio world, we've used em for low-band amplification for low-power transmitters (known as QRP rigs) with a lot of success, and if u blow one up, it's $3.00 to replace the final . They work up to about 20MHz with no problems (usually ), but seem to crap out with unity gain anywhere much above that frequency.

Randy

[nigelwright7557]

I have used them in class AB and class D.
They work very well.

[AB5NI]

Originally Posted by nigelwright7557

I have used them in class AB and class D.
They work very well.

Nice to know. Going to play around with it in a couple of weekends and see what gives.

Thanks for the info,

Randy

[nigelwright7557]

Originally Posted by AB5NI

Nice to know. Going to play around with it in a couple of weekends and see what gives.

Thanks for the info,

Randy

Class AB is a little more forgiving for beginners than class d.
Class d is a minefield for beginners, it is fussy about pcb layout and decoupling.

There are quite a few threads on here with various class AB amps.
Beware of some of those published on internet sites, some are poor and dont work reliably.

[AB5NI]

Originally Posted by nigelwright7557

Class AB is a little more forgiving for beginners than class d.
Class d is a minefield for beginners, it is fussy about pcb layout and decoupling.

Well, I'm pretty much consider myself a beginner in electronics, although I have studied the subject for many years . (Read: I'm an extra class ham, so I know a bit about RF, and I've designed and built various transmitters and receivers, but I'm no RF engineer .) I guess the analog RF work will somewhat apply to audio, though. Just have to drop the frequency range a bit .

As far as AB vs. D goes, I'll definitely stick with AB. OTOH, let's say that I did go class D and would wound up with some form of parasitic oscillation and the MOSFET cratered...not a really big deal, right? Just a few bucks for a new FET and we're off and trying again .

As a matter of fact, since power MOSFET's are so damn cheap, you'd think you'd see a lot more designs out there that took adavantage of that situation. Maybe something like a "Radio Shack Special" for beginners or something, yet I haven't located such a circuit in my few days of searching. One would tend to think that this circuit would be prevalent, especially in this day and age and state of the global economy. Go figure .

There are quite a few threads on here with various class AB amps.
Beware of some of those published on internet sites, some are poor and dont work reliably.

Oh, I know this only too well from trying to replicate the circuits in the ham mags over the years . What I tend to look for are circuits that others have replicated, with forum posts that mention it was an easy build and did exactly what an article said it would do. That, and I tend to gravitate toward circuits that are so well liked that a forum member decided to make up a PCB and is doing a group-sell on the thing at little or no profit. That, in my humble opinion, speaks volumes for the circuit. (Pun intended .)

Anywho, thanks for the great info, Nigel!

Randy

[CBS240]

Hi

When it comes to using power mosfets as the output stage for linear amplifiers, there are two general schools of thought. One is to naively use them straight forward as if they are BJT like devices, and the other is to use them like they are mosfets....with all their inherent issues, and advantages. I tend to go with the latter, they physically work on an entirely different principle than BJTs. You can grossly over-dampen a mosfet and keep it from oscillating, adding forgiveness to layout and local compensation techniques, but you end up throwing away many of the features, namely superb Ft and phase margin, that power mosfets offer over power BJTs, and I dare to mention the Gm issues regarding class AB bias. Some of the obvious issues include non-linear capacitance, gate charge requirements for majority carrier motion to change conduction (inversely proportional to frequency), parasitic RF components that can cause instabilities and hair-pulling moments when the circuit self-destructs for no apparent reason. It happens when you don't take into consideration all the angles, RF damping, PCB layout, ect.

To simply state MHO, mosfets can make better output transistors than BJT's, but it is very easy to make a poor or average performing mosfet amplifier. Don't think because of an apparent high input capacitance that a high BW very low distortion amplifier can't be built using mosfets, the last proto-amp I made has wide BW, -3dB gain at ~500KHz. Many see no advantage to being able to amplify at full gain a 200KHz or even 500KHz sine wave, my ears regarding sound stage perception say otherwise. BJT's are just much more forgiving when it comes to parasitics. Conversely mosfets are much more forgiving in temperature coefficient - thermal bias drift and SOA. There are several good threads on mosfet amplifiers in the forums. Basically there are two types of power mosfets, referred to as Vertical and Lateral. They have differing properties such as Gm and zero-temperature coefficient currents, but also very differing transfer functions, and a difference in price and availability. These things need to be considered when deciding on which technology to use. Laterals are created with the intention of linear use and are the simplest to implement. Their transfer characteristics very much resembles the Square Law. Verticals are created for switching and with a more exponential transfer, are not designed for linear use in mind, however certain types of verticals seem to excel quite well for linear power use if used properly.

[AB5NI]

Howdy, CBS240, and thanks for the reply.

Originally Posted by CBS240

Hi

When it comes to using power mosfets as the output stage for linear amplifiers, there are two general schools of thought. One is to naively use them straight forward as if they are BJT like devices, and the other is to use them like they are mosfets....with all their inherent issues, and advantages. I tend to go with the latter, they physically work on an entirely different principle than BJTs. You can grossly over-dampen a mosfet and keep it from oscillating, adding forgiveness to layout and local compensation techniques, but you end up throwing away many of the features, namely superb Ft and phase margin, that power mosfets offer over power BJTs, and I dare to mention the Gm issues regarding class AB bias.

Speaking of Ft, in the ham world, we've had em up into the 20-Mhz with few issues, although like you've stated, we've had to dampen the circuity to hinder oscillation for ease of replication of the circuit. "In the ballpark" is good enough for our uses, but that's the RF-Ham world. I can see that in high-end audio reproduction (total noob in this area), such things wouldn't be the norm, where overall linearity and distortion are the main factors at hand. The problem there, it would seem, would be that circuit replication becomes a bit more difficult, right? (BTW, we mainly do class-C operation with our power MOSFET's, which gives us a bit of a boost in CW and FM, although there are some brave souls out there that have "tried" AB with a bit of success, but it's usually crater-city for the FET, especially in high SWR situations . Not too much of a big issue, though -- a couple of bucks and you're back on the air .)

Some of the obvious issues include non-linear capacitance, gate charge requirements for majority carrier motion to change conduction (inversely proportional to frequency), parasitic RF components that can cause instabilities and hair-pulling moments when the circuit self-destructs for no apparent reason. It happens when you don't take into consideration all the angles, RF damping, PCB layout, ect.

Oh, I know these issues only too well, unfortunately . Playing around with FETs in the VHF region should be experienced by everyone at least once or twice in their lives . And may the Lord have mercy on ur soul if u crater a GASFET up in that region of the spectrum .

To simply state MHO, mosfets can make better output transistors than BJT's, but it is very easy to make a poor or average performing mosfet amplifier. Don't think because of an apparent high input capacitance that a high BW very low distortion amplifier can't be built using mosfets, the last proto-amp I made has wide BW, -3dB gain at ~500KHz. Many see no advantage to being able to amplify at full gain a 200KHz or even 500KHz sine wave, my ears regarding sound stage perception say otherwise.

Well, I remember reading something many years ago about this, CBS, and the conclusion that was reached was this: you might not hear a difference, but your body will definitely feel the difference. Had something to do with body capacitance and the hairs on your arms resonating with fundamental frequencies and mainly harmonics. Basically, these frequencies caused goosebumps on peoples bodies, and they rather liked that situation.
So, by all means, please keep up with the broadband circuity. I, for one, will definitely build one of your amps or would consider a purchase of one of your designs on your love of broadband design alone.

BJT's are just much more forgiving when it comes to parasitics. Conversely mosfets are much more forgiving in temperature coefficient - thermal bias drift and SOA.

Which is the main reason we like FETs in VFO design of our more simple radios. The latest trend we've embraced is the AD9851 DDS chip, which allows our radios to have VFO operations from the 0-60MHz range. Lots of fun, and you get a general-coverage receiver/transmitter to boot .

There are several good threads on mosfet amplifiers in the forums. Basically there are two types of power mosfets, referred to as Vertical and Lateral. They have differing properties such as Gm and zero-temperature coefficient currents, but also very differing transfer functions, and a difference in price and availability. These things need to be considered when deciding on which technology to use. Laterals are created with the intention of linear use and are the simplest to implement. Their transfer characteristics very much resembles the Square Law. Verticals are created for switching and with a more exponential transfer, are not designed for linear use in mind, however certain types of verticals seem to excel quite well for linear power use if used properly.

Ah. Nice to know. I'll make sure I read up on the subjects and peruse the info, CBS, and thanks for the informative reply!

More like $100.00 of info for me

Randy

[CBS240]

Hi

If you used a power fet for class C, I suspect you would use a vertical type. They would have the greater Ft of the two due to the overall lesser Gm of the lateral type device, the lateral's available Ft is diminished a little. Also class C is <180 degrees and is more like a switch than class A-AB. 'Vertical' is a general term to discribe a group of different devices from fabrication processes that lead to more exponential transfer. There are different types of vertical fets such as Hex-fet, Trench-fet, U-fet, Uni-fet, plainer stripe-fet, and some new processes I've not heard of I'm sure. Although all these technologies were developed for switching not all of these types are suitable for linear power. Trench and U types seem to 'hot spot' like BJTs, something like a secondary breakdown. Hex-fet and plainer stripe seem to handle higher Vds SOA better, no secondary breakdown. These power fets are very popular for class A amplifiers because of their properties and the fact you don't have the Q-point anywhere near the low Id region where Gm drops significantly, and so the transconductance remains constant. This leads to much less distortion created by the device. Class A uses and wastes lots of energy. (unless you use a tracking power supply for the output stage, but that opens up a whole bucket of worms.) Class AB is a good compromise but requires operation of the fets across the low Id region, and this creates issues that have to be addressed. There have been some clever ways developed that try to 'linearize' Gm wrt Id for all 360 degrees. Since we don't live in a perfect world we have to compromise on trade offs in order to reach the point closest to the goals set. Audio is just another branch of electronics and not being a newbie to electronics you already know this.
One of the difficult things to achieve in audio power amps is stability. The circuit has to drive a load that varies with respect to frequency in magnitude up to 30%, phase angle, lead or lag, i e reflecting energy back into the amplifier, I suppose kind of like a SWR (I don't consider myself a RF guy tho ), and not crater the fets, and operate 360 degrees with constant linear voltage gain across a wide band of frequencies. Also there is power supply rejection and noise to consider, and that goes for each stage of the amplifier.

BTW, I recommend Bob's book, it is a great technical reference and is easy to understand if you have the basic background.

[AB5NI]

Originally Posted by CBS240

Hi

If you used a power fet for class C, I suspect you would use a vertical type.

Yep. Sure do. Most of the designs for low-power amplification are single-ended and based around the IRF510 series Hex-Fet. $3.00-$4.00 at Radio Shack . A good example would be NB6M's "Mini Boots" found here:
NorCal QRP Club - NB6M Mini Boots


Bunch of great info deleted for brevity, and thanks a ton for spending time explaining this stuff to a noob, CBS! ...

Class A uses and wastes lots of energy. (unless you use a tracking power supply for the output stage, but that opens up a whole bucket of worms.)

Yes, Class A is a bit inefficient, but does it sound oh so sweet in the audio region . A few years back, I was working at an engineering firm in Houston,
and I ran across a physicist doing computational fluid dynamics for some designs we were working on. Anywho, he was also into audio in a big way, and we got to talking about class A amplification vs. AB, mono-blocks vs. integrated amps, and companies like Audio Research and whatnot. We decide to go over to Audio Research's site and see what was the latest, greatest thing coming out of their facilities, and we run across the ever-popular and plentiful 811A amplifier, biased class A. I told Paul (the CFD dude), "Man, it looks like they're getting 17.5 watts out of that 811A, and I'm wondering how in the heck they're keeping the grids from shorting with all that heat being generated . The supply was built like a tank, of course, and I'd imagine everything was mil-spec and rated at 100% duty cycle, continuous commercial service and then some!

To make a long story short here: for me, I guess I'm a purist when it comes to audio. I'd much prefer class-A, single-ended designs at low wattage than push-push or push-pull designs. They are a lot less complicated (in most respects), no need for ballast resistors and all the other complications needed for high-power output, and let's not forget trying to impedance match that kinda setup to a varying, low-impedance load . So, "give me single-ended class A or give me death!" . BTW, do u guys use a lot of L-Network theory in this regard or delve more in the transmission-line area? I would imagine it's more L-Network when dealing with auido transformers and all. (An inquiring noob would like to know! .)

Class AB is a good compromise but requires operation of the fets across the low Id region, and this creates issues that have to be addressed. There have been some clever ways developed that try to 'linearize' Gm wrt Id for all 360 degrees. Since we don't live in a perfect world we have to compromise on trade offs in order to reach the point closest to the goals set. Audio is just another branch of electronics and not being a newbie to electronics you already know this.

Yeah, I know it, but that doesn't mean I havta like it, does it? . Seriously, I definitely see the reasons why u guys do things like AB and push-pull and what have you, especially when you're dealing with design constraints and costs. I'm all about compromise -- that's for sure. Hell, I design and build radios out of junk parts stripped from old TV sets, computers, and the like. I'm all about compromise, in most situations. Well, I do tend to put my foot down when designing an old-school VFO. Disc-ceramics just don't cut it in a VFO (drift due to temperature variations), so I put my foot down, get on the net, and purchase polystyrene and NP0 caps like you are supposed to do . Wait. Let me guess. You guys are all about polystyrene, NP0, Mylar, and nylon, and anything else in the audio circuitry is just plain noob city .

One of the difficult things to achieve in audio power amps is stability. The circuit has to drive a load that varies with respect to frequency in magnitude up to 30%, phase angle, lead or lag, i e reflecting energy back into the amplifier, I suppose kind of like a SWR (I don't consider myself a RF guy tho ), and not crater the fets, and operate 360 degrees with constant linear voltage gain across a wide band of frequencies.

Yep. We don't have to worry about most of this, although a solid, regulated constant-current supply on the frequency-determining components (such as a VFO) and the audio chain are a must, especially when you consider having 90db gain in the audio amp and such in our more simple receivers. You also have to worry about not over coupling the VFO output, 12V (usually) rail being properly bypassed at audio and RF and such. A lot of little nits to pick, to say the least, and I'm sure it's the same in the audio region of the spectrum .

Oh, and on SWR. A plus. Step to the head of RF class, CBS .

Also there is power supply rejection and noise to consider, and that goes for each stage of the amplifier.

Same worries we have at RF, CBS, except now you have worry about stuff like 3rd-order harmonics and the like finding their way into the audio chain, and here you are with 90-100db of amplification, listening to squeals and squaks that have certainly ruined your ears to anything around the 3-5Khz region . Anywho, we do have these worries -- and then some -- and it's really irritating when you see three-pole regulars not properly bypassed to make minimalistic radios with a low parts count and low-current draw. A lot of the low-power crowd tend to be the outdoors type, bringing their radios along on jaunts out into the woods, fishing, etc., so current draw is almost always paramount in the design. Other design goals and worries would be: IMD, MDS (minimum detectable signal), 3rd order intercept, and noise floor, but you usually see those kinda worries in our "single-ended, mono-block, class A" like radio designs. (Read: our expensive as crap radios .)

BTW, I recommend Bob's book, it is a great technical reference and is easy to understand if you have the basic background.

Well, I'm hoping that I'll hop on low end of Bob's design goal for the book. I'm okay with the math and physics..no problem there. It's just the lingo that's probably going to trip me up a bit, I'm sure, but I'm sure I'll find some other book I have or maybe even the net (e.g., wikipedia) to help explain things.

Nice talking to you, CBS....

Randy