I usually sim amplifiers without LR network.
Good amp (except some cutting edge, high-bandwidth ones) should be stable and produce decent square waves WITHOUT it.
If I feel it's needed, I would add LR after the sim (and re-sim with it).
I feel that if you have a simple amp, and you have ringing, adding LR to solve the problem is not the best way....
Good amp (except some cutting edge, high-bandwidth ones) should be stable and produce decent square waves WITHOUT it.
If I feel it's needed, I would add LR after the sim (and re-sim with it).
I feel that if you have a simple amp, and you have ringing, adding LR to solve the problem is not the best way....
I removed the coil and ran a few sims on various reactive loads. Clean squares and plenty of phase margin even without the LR. The proof is of course in the physical version. I'll go with the external-if-needed approach for now, I think.
Doing a few final sanity checks, then PCB layout. I'm excited to listen to this thing!
Doing a few final sanity checks, then PCB layout. I'm excited to listen to this thing!
Final(?) cleaned-up schematic. Looks very Bog Standard and Blameless-like to me. Remember that the aim was for something that's easy to build from bog standard components. Originality was never a goal.
PCB layout is underway.
PCB layout is underway.
First draft of a PCB. This is definitely not my forte, but I'm learning. I tried to keep all high current stuff on the bottom and I'm using some kind of star ground. I'll post the actual layout for you to pick apart once I've cleaned it up.
Yes, those are spade connectors. I like them for high current stuff. But maybe you have a better suggestion?
Yes, those are spade connectors. I like them for high current stuff. But maybe you have a better suggestion?
Q14 is upside down, EC swapped. Q4 and Q6 could use a base resistor or CE capacitor to guarantee CCS stability, and/or use something cheap like 2n3906 instead of 2n5401. Using the same part with no resistor or cap in such a CCS can be unstable. R3,R5 may be a bit high, too much drop. R4 could be maybe 0.65/2mA ~= 330.
Oh boy! You just saved me a ton of cursing. Thank you!!!!Q14 is upside down
As for the LTP load resistors, 470ohm gave me the lowest distortion in simulations, but I will definitely try lower values in the physical build.
Regarding these two CCSes - you could split R14 and R19 into two halves, and add capacitor from the junction to the rail.
See example below.
This is good for PSRR... I think you have enough space on the PCB left...
See example below.
This is good for PSRR... I think you have enough space on the PCB left...
Great minds think alike! 😀 I was just thinking of doing that after leafing through Self’s book over a beer last night.
Rocks think alike too! 🙂....thanks for the reminder... I need to go get more beer.... er... should have said precious stones...
Pressed for space, the second half of the resistor and cap can be shared between two (or four) CCS’s.
I hope you don't take me too seriously njswede.
My friend George also thinks like I do. He doesn't speak much... and really hasn't spoken much in years now that I come to recollect. He has a great mind... or should I say more correctly head... being reincarnated as a lettuce.
My friend George also thinks like I do. He doesn't speak much... and really hasn't spoken much in years now that I come to recollect. He has a great mind... or should I say more correctly head... being reincarnated as a lettuce.
The bias threshold of these FETs is about 4V each, so you lose about 10VPP output swing vs rail-rail, which is more of an issue when running a lower supply voltage. There are ways to avoid this but then it's not so "bog standard". Do you care?
In my particular build, it not going to matter, since the supply produces about 37V per rail and I’m only planning to pull 50W. But I’m interested in your suggestion nonetheless!
There are several variations on bootstrapping. The drivers can get in the way, but you can bootstrap the driver collectors. Bob C is adamant about the need for drivers, but I think that applies to "high performance" high slew rate amps. The other option is a CFP output with gain. I have a "fix" for a BJT CFP output but it's probably not an issue for FETs. I could dig up some simulations if you like.
Simulations seem to suggest that you're right, at least for a little 50W job like this one. The current source on top of the VAS seems to have enough oomph to drive the MOSFETs. The slew rate DID go from 13V/us to 19V/us when I added the drivers. Not sure how audible that is, though. It was mostly because people on this thread were telling me the sky was going to fall of I didn't (well, not really, but almost 😊).There are several variations on bootstrapping. The drivers can get in the way, but you can bootstrap the driver collectors. Bob C is adamant about the need for drivers, but I think that applies to "high performance" high slew rate amps.
Speaking about the CCS current: Do you want to run them that hot? I make Q8 and Q16 about 550mW on a 625mW max part. Not a lot of margin. I would do the layout to accept a BD139+BD140 (etc) as an option, and/or leave room for a small heat sink.
You could do a partial bootstrap for everything ahead of the FETs (4V/40V=10%). That would be very minimal, ie a resistor, assuming you need some local filters anyway. And no-load/ jumper if you choose not to.
You could do a partial bootstrap for everything ahead of the FETs (4V/40V=10%). That would be very minimal, ie a resistor, assuming you need some local filters anyway. And no-load/ jumper if you choose not to.
I calculated something like 488 mW, the voltage drop across the VBE multiplier helps a bit. That is below the rating up to 52.4 degrees Celsius, but taking no margin for signal-related temperature fluctuations. I also didn't take into account that the current drops somewhat at high ambient temperatures, and I guessed the base-emitter voltage.
It's not nearly as bad as what I have seen in a commercial amplifier design from Philips, but it would certainly be good for reliability to reduce the current a bit or improve cooling.
It's not nearly as bad as what I have seen in a commercial amplifier design from Philips, but it would certainly be good for reliability to reduce the current a bit or improve cooling.
- Home
- Amplifiers
- Solid State
- "The Bog Standard" - A good enough amplifier for the rest of us