Well... Here goes nothing!
I've been reluctant to do this out of fear of ridicule. You see, I'm not a very experienced designer and the design I'm about to present is a pretty mundane Class AB job.
But lately I've seen beginners on various forums struggle with designs that require exotic and/or obsolete components or designs that require very stable power supplies and careful thermal management. I've also seen people having a tough time with what's clearly outdated or outright bad designs that are floating around. So based on that, I thought a fellow beginner might find this thread useful.
I decided to share a decent amplifier that can be built using cheap components that are readily available from almost any source. It uses only 2N5551/5401, BD139 and IRFP(9)240 transistors, some 1N4148 diodes, a couple of 5V zeners and a smattering of capacitors and resistors. That's it! It runs fairly cool, does fantastic with a modest 40mA bias current and has a net negative temperature coefficient with regard to bias if you're careful with the thermal transfer between the VBE multiplier and the output MOSFETs. It physical tests, it's fairly flat past 100kHz at 50% power and gets a decent 0.003% THD at 80% power and 1kHz in simulations. I don't have the equipment to measure an accurate physical THD, but just eyeballing it on the scope, it's definitely <0.1% and probably a lot less. Nothing compared to the Class A masterpieces some people here are building, but enough to sound pretty nice. The way it's drawn it should be good up to 50W with the right heatsink.
It's also very forgiving in terms of component values and (probably) layout. I've messed around in SPICE, and it takes real work to make it misbehave. You can even mess up the matching of the LTP transistors quite badly without it throwing a fit. It also has a >70 degree phase margin in simulations and would probably do fine with a lower Miller capacitance if someone wants faster performance. I've tried it with 8ohm and 6ohm speakers, and based on simulations it should have no problem at all with 4ohm as long as you're staying within a reasonable power (50W). The gain is 23x (27dB), which is a bit low for regular line level. You could probably bump it up a bit, but I'd probably recommend a pre-amp instead. I haven't checked the stability with a higher gain.
This design has been built and sounds and measures well. Audiophile quality? Probably not, but good enough to "open up" music I've listened to many times before and letting me discover new details.
I'll share the schematic and also a Github repo with all the design and simulation files, along with some ideas on how to build and test it. If you're a beginner, feel free to try it or just borrow ideas and inspiration. And if you're an experienced designer, I'm happy to listen to ideas for improvement. You can even clone the repo and send me a pull request if you want to contribute. But please be gentle! 🙂
There are a few things I'd like to change when I get a chance. The pot on the long tail can be replaced with fixed resistors and the terminal block for the speakers on the middle of the board is awkward. But it's buildable and works well in its current form. You could (and probably should) add speaker protection and maybe a preamp and some tone controls.
The repo: https://github.com/prydin/classab-amp-mosfet
I hope you'll find my humble contribution valuable in some way. And be gentle on me!
[EDIT: Updated the schematic with a protection diode for the VAS to avoid disasters... Also removed the DC offset pot, since it doesn't do much due to the current mirror]
I've been reluctant to do this out of fear of ridicule. You see, I'm not a very experienced designer and the design I'm about to present is a pretty mundane Class AB job.
But lately I've seen beginners on various forums struggle with designs that require exotic and/or obsolete components or designs that require very stable power supplies and careful thermal management. I've also seen people having a tough time with what's clearly outdated or outright bad designs that are floating around. So based on that, I thought a fellow beginner might find this thread useful.
I decided to share a decent amplifier that can be built using cheap components that are readily available from almost any source. It uses only 2N5551/5401, BD139 and IRFP(9)240 transistors, some 1N4148 diodes, a couple of 5V zeners and a smattering of capacitors and resistors. That's it! It runs fairly cool, does fantastic with a modest 40mA bias current and has a net negative temperature coefficient with regard to bias if you're careful with the thermal transfer between the VBE multiplier and the output MOSFETs. It physical tests, it's fairly flat past 100kHz at 50% power and gets a decent 0.003% THD at 80% power and 1kHz in simulations. I don't have the equipment to measure an accurate physical THD, but just eyeballing it on the scope, it's definitely <0.1% and probably a lot less. Nothing compared to the Class A masterpieces some people here are building, but enough to sound pretty nice. The way it's drawn it should be good up to 50W with the right heatsink.
It's also very forgiving in terms of component values and (probably) layout. I've messed around in SPICE, and it takes real work to make it misbehave. You can even mess up the matching of the LTP transistors quite badly without it throwing a fit. It also has a >70 degree phase margin in simulations and would probably do fine with a lower Miller capacitance if someone wants faster performance. I've tried it with 8ohm and 6ohm speakers, and based on simulations it should have no problem at all with 4ohm as long as you're staying within a reasonable power (50W). The gain is 23x (27dB), which is a bit low for regular line level. You could probably bump it up a bit, but I'd probably recommend a pre-amp instead. I haven't checked the stability with a higher gain.
This design has been built and sounds and measures well. Audiophile quality? Probably not, but good enough to "open up" music I've listened to many times before and letting me discover new details.
I'll share the schematic and also a Github repo with all the design and simulation files, along with some ideas on how to build and test it. If you're a beginner, feel free to try it or just borrow ideas and inspiration. And if you're an experienced designer, I'm happy to listen to ideas for improvement. You can even clone the repo and send me a pull request if you want to contribute. But please be gentle! 🙂
There are a few things I'd like to change when I get a chance. The pot on the long tail can be replaced with fixed resistors and the terminal block for the speakers on the middle of the board is awkward. But it's buildable and works well in its current form. You could (and probably should) add speaker protection and maybe a preamp and some tone controls.
The repo: https://github.com/prydin/classab-amp-mosfet
I hope you'll find my humble contribution valuable in some way. And be gentle on me!
[EDIT: Updated the schematic with a protection diode for the VAS to avoid disasters... Also removed the DC offset pot, since it doesn't do much due to the current mirror]
Last edited:
Few observations after quick 1st look:
1) You don't need D8/D9. They are built-in in IRFPs
2) If using only 1 pair of output devices you don't need source resistors. Their absence will lower Thd
3) From my experience, hexfets without drivers sound dreadful, even though amp simulates OK
Latfets without drivers sound OK.
4) To accurately track temperature of hexfets, Vbe multiplier Q7 may need extra diode in the emitter (or another bjt)
Temperature in LtSpice simulation would reveal if temp tracking works as expected.
1) You don't need D8/D9. They are built-in in IRFPs
2) If using only 1 pair of output devices you don't need source resistors. Their absence will lower Thd
3) From my experience, hexfets without drivers sound dreadful, even though amp simulates OK
Latfets without drivers sound OK.
4) To accurately track temperature of hexfets, Vbe multiplier Q7 may need extra diode in the emitter (or another bjt)
Temperature in LtSpice simulation would reveal if temp tracking works as expected.
Thank you!
1. I’ve been told never to trust the body diode for protection. But you’re probably right. Then again, it’s 5 cent diode.
2. Interesting. I was told by someone that I should INCREASE the value of the source resistors. I’ll certainly experiment with that. Thanks!
3. Dreadful how? It measures fine as far as I can tell and sounds good too. What aspect of the sound goes bad? THD? Frequency response? The only thing I may have noticed that the string section in large orchestras sound a little “metallic”, but that could be the room or the decent but not great Dayton towers I have it hooked up to at the moment.
4. With the VBE multiplier in close thermal contact with the MOSFETs I have a negative temperature coefficient. If I listen at a high volume for a minute, the bias current is lower than before I cranked up the volume by about 25%.
1. I’ve been told never to trust the body diode for protection. But you’re probably right. Then again, it’s 5 cent diode.
2. Interesting. I was told by someone that I should INCREASE the value of the source resistors. I’ll certainly experiment with that. Thanks!
3. Dreadful how? It measures fine as far as I can tell and sounds good too. What aspect of the sound goes bad? THD? Frequency response? The only thing I may have noticed that the string section in large orchestras sound a little “metallic”, but that could be the room or the decent but not great Dayton towers I have it hooked up to at the moment.
4. With the VBE multiplier in close thermal contact with the MOSFETs I have a negative temperature coefficient. If I listen at a high volume for a minute, the bias current is lower than before I cranked up the volume by about 25%.
I was just glancing over Bob Cordell's Designing Audio Power Amplifiers book. Easy to find on the web. In Chapter 12.3 - Error Correction for MOSFET Output Stages, he shows a simple schematic of an amplifier using the IRFP240/9240 and pre-drivers.
I have nothing useful to add but that statement!
Best,
Jose
I have nothing useful to add but that statement!
Best,
Jose
interesting. However, in simulations, the high OLG global feedback seems to do a pretty good job at canceling any non-linearities in the output transistors.I was just glancing over Bob Cordell's Designing Audio Power Amplifiers book. Easy to find on the web. In Chapter 12.3 - Error Correction for MOSFET Output Stages, he shows a simple schematic of an amplifier using the IRFP240/9240 and pre-drivers.
I have nothing useful to add but that statement!
I guess I need to read that entire chapter.
I was not making any value judgments on the merits of pre-driver vs no pre-driver, as I am not smart enough for that sort of thing. Best to leave it to experts like Bob.
He did mention that "While the power MOSFET has many advantages, it was pointed out in Chapter 11 that the lower transconductance of the MOSFET will result in moderate crossover distortion unless rather high bias currents are chosen"
what's the bias for the FET's?
Also, being cheap, I appreciate anyone willing to put forth designs that cater to that!
Best,
Jose
He did mention that "While the power MOSFET has many advantages, it was pointed out in Chapter 11 that the lower transconductance of the MOSFET will result in moderate crossover distortion unless rather high bias currents are chosen"
what's the bias for the FET's?
Also, being cheap, I appreciate anyone willing to put forth designs that cater to that!
Best,
Jose
The bias current, from what I can remember is about 100mA (I know I said 40mA in the original post, but I think I might have been mistaken. I’ll check tomorrow)
Coefficient is negative just BJT tend to over track bias, it will lower bias to much.With the VBE multiplier in close thermal contact with the MOSFETs I have a negative temperature coefficient. If I listen at a high volume for a minute, the bias current is lower than before I cranked up the volume by about 25%.
The "non linearity's " are only at high frequency so people set the bias pretty high to overcome any visible crossover distortion.
They are dirt easy to drive, but the non linear behavior at high frequency needs usually 20ma of current.
So that is what you will find with low distortion designs, really high current to drive them and really high bias current.
Otherwise for a basic amplifier and distortion people pretend to hear isnt actually there unless measured or your shooting for high frequency bragging rights.
Maybe noticeable sound difference with T0-126 drivers biased high around 15ma then bias the Mosfets to 200ma or 300ma. Easier to use BJT at that point.
Excluding voltage losses to the rail swing.
There is more advanced circuits to thermal track with BJT and not over track. If you wanna weasel it have the thermal tracking transistor close to the heatsink using a TO-92 plastic type. Or if direct mounted with To-126 can weasel it by stacking more mica insulators so it is not as sensitive
R8 - R18 dont need to be 330 ohms, that is typical oscillator lateral mosfet values, they can be 33 to 47 ohms.
Typical gate protection diodes used, will behave like current limiters, so if they grab during operation with 4 ohm or 3 ohm loads you get flyback
so its why you see typical flyback diodes on the outputs like you show. Yes they have built in diodes but people often add them anyways for assurance.
Usually expect about 5x or more flyback than the rails so be typical to rate the diode with 5x even 10x voltage rating.
Seems the Zener's are D6/D11 at the gates around 5.6 volts. So people often use a looser value since with 4 or 2 ohm loads its likely they will clamp early and cause flyback. So often set to max value or close to max value like 10 volts. Would have to test in sim, people like the current limit if the amp is shorted.
Maybe 8 volts is a good tradeoff, easy to test in sim.
I'm more the audience for a project like this than I am someone who could hope to contribute, but I just wanted to say that I think that what you're shooting for is something really needed in the DIY amplifier market.
While there are plenty of great Class A designs to choose from (thanks Nelson), and LM3886 chip amp designs abound, a simple, 'good enough' discrete AB amp design for beginners to try to build and understand is not something I've really seen a lot of. Discrete AB DIY options for which you can buy PCBs seem to be either hardcore SINAD-banishing high-wattage uber-projects, interesting but eccentric curiosities, or derelict designs due to some kind of unobtainium or lack of support.
After I get to the end of the Class A amp I'm building (and recover financially 😅 ), this is the sort of thing I'll be looking to build. 🙂
While there are plenty of great Class A designs to choose from (thanks Nelson), and LM3886 chip amp designs abound, a simple, 'good enough' discrete AB amp design for beginners to try to build and understand is not something I've really seen a lot of. Discrete AB DIY options for which you can buy PCBs seem to be either hardcore SINAD-banishing high-wattage uber-projects, interesting but eccentric curiosities, or derelict designs due to some kind of unobtainium or lack of support.
After I get to the end of the Class A amp I'm building (and recover financially 😅 ), this is the sort of thing I'll be looking to build. 🙂
Bravo for the initiative. Please consider C1 to be a smaller value film cap like 4.7 µF. C4 a bipolar MUSE ES with 100 nF film in parallel. 475 kOhm over pins 1 and 2 of J3. R1 1 kOhm and C2 1000 pF. RC filters in the rail voltages to the VAS would not hurt either.
Last edited:
2. Interesting. I was told by someone that I should INCREASE the value of the source resistors. I’ll certainly experiment with that. Thanks!
Not for mosfets. One pair of mosfets does not need source resistors. Search this forum, this was discussed several times (E.g. by Ian Hegglun).
3. Dreadful how? It measures fine as far as I can tell and sounds good too. What aspect of the sound goes bad? THD? Frequency response? The only thing I may have noticed that the string section in large orchestras sound a little “metallic”, but that could be the room or the decent but not great Dayton towers I have it hooked up to at the moment.
I can't measure it or sim it. All I know, that when I tried to build hexfet amps without drivers, they sounded terrible.
If drivers were "optional", I'm sure commercial companies/designers would be skipping them to save money, but they don't.
You'll rarely see hexfet amp without drivers.
4. With the VBE multiplier in close thermal contact with the MOSFETs I have a negative temperature coefficient. If I listen at a high volume for a minute, the bias current is lower than before I cranked up the volume by about 25%.
I'm sure you have negative coefficient. The question is 'how accurate' ? Cordell has a whole chapter on Vbe multipliers in his book, including cases for hexfet outputs. Thermal tracking for BJTs and mosfets is not the same, they are different.
There is several ways to implement Vbe multiplier for them, and have as good as possible thermal tracking.
Single transistor will work, but it's not perfect. The easiest way to improve it, is to add a diode (some people use red LED) in the emitter of the tracking transistor.
Again - you can find plenty of hexfet amps showing this. e.g. here:
Not for mosfets. One pair of mosfets does not need source resistors. Search this forum, this was discussed several times (E.g. by Ian Hegglun).
3. Dreadful how? It measures fine as far as I can tell and sounds good too. What aspect of the sound goes bad? THD? Frequency response? The only thing I may have noticed that the string section in large orchestras sound a little “metallic”, but that could be the room or the decent but not great Dayton towers I have it hooked up to at the moment.
I can't measure it or sim it. All I know, that when I tried to build hexfet amps without drivers, they sounded terrible.
If drivers were "optional", I'm sure commercial companies/designers would be skipping them to save money, but they don't.
You'll rarely see hexfet amp without drivers.
4. With the VBE multiplier in close thermal contact with the MOSFETs I have a negative temperature coefficient. If I listen at a high volume for a minute, the bias current is lower than before I cranked up the volume by about 25%.
I'm sure you have negative coefficient. The question is 'how accurate' ? Cordell has a whole chapter on Vbe multipliers in his book, including cases for hexfet outputs. Thermal tracking for BJTs and mosfets is not the same, they are different.
There is several ways to implement Vbe multiplier for them, and have as good as possible thermal tracking.
Single transistor will work, but it's not perfect. The easiest way to improve it, is to add a diode (some people use red LED) in the emitter of the tracking transistor.
Again - you can find plenty of hexfet amps showing this. e.g. here:
Amp from the post https://www.diyaudio.com/community/threads/unusual-amp-from-1987.357369/post-7068145
has been built (1 channel so far). Works like a champ. the only correction was to change C19 from 5pF to 9pF.
One channel built and tested so far. Will build 2nd channel, and then test with real music.
Idle current 40-50mA per output fet, output DC voltage (with C20 installed): 0.4 mV. Without out - 160mV.
has been built (1 channel so far). Works like a champ. the only correction was to change C19 from 5pF to 9pF.
One channel built and tested so far. Will build 2nd channel, and then test with real music.
Idle current 40-50mA per output fet, output DC voltage (with C20 installed): 0.4 mV. Without out - 160mV.
Last edited:
I was told to increase the source resistors by someone who also said current mirrors in the LTP was bad, so… 😀 I removed the resistors in SPICE and it didn’t make any difference. Next time I open the case, I’ll just try to short them out and see (or rather hear) what happens.
I’ll also run some simulations with drivers. Just a simple push-pull pair, right?
Thanks for your feedback!
I’ll also run some simulations with drivers. Just a simple push-pull pair, right?
Thanks for your feedback!
Actually, since you have no drivers, thermal tracking Vbe multiplier should be designed specifically for this case. All typical cases
for hexfets described in the literature (Cordell, Self), assume that drivers are present.
If you have drivers, a diode in the emitter of the tracking device is good enough..
LTSpice has something called "Thermal simulation" where you can specify temperature of each device, and see how this affects currents/voltages.
But I think that the best improvement to this amp, that would actually make sound better, will be adding drivers.
Changing Vbe multiplier is secondary. If you are not planning to run your amp at the max power, it will work fine.
yes.
for hexfets described in the literature (Cordell, Self), assume that drivers are present.
If you have drivers, a diode in the emitter of the tracking device is good enough..
LTSpice has something called "Thermal simulation" where you can specify temperature of each device, and see how this affects currents/voltages.
But I think that the best improvement to this amp, that would actually make sound better, will be adding drivers.
Changing Vbe multiplier is secondary. If you are not planning to run your amp at the max power, it will work fine.
I’ll also run some simulations with drivers. Just a simple push-pull pair, right?
yes.
Yeah, the Miller cap might be a tad large. But slew rate didn’t seem to be an issue, so I left it as it is and enjoyed the peace of mind of lots of phase margin instead. 😀
+1But I think that the best improvement to this amp, that would actually make sound better, will be adding drivers.
Last edited:
- Home
- Amplifiers
- Solid State
- "The Bog Standard" - A good enough amplifier for the rest of us