In that case you've probably exceeded the performance of 99% of the other EF2s out there in the world and a large % of all other amps too. At least as far as distortion is concerned, but the implementation is important and you've obviously got that down, which a lot do not. And of course the good implementation is necessary to illustrate the small changes that you're making.
The only areas where I'd think you'd get a better design might be in using a triple EF to reduce the VAS loading and to use better outputs like the mjl4281, reliability should improve as would the ability to drive tougher loads, with possibly fewer output pairs (providing it's stable!). If the heatsinking allowed and was a direction you'd like to move in, you could reduce the value of the RE resistors to increase the optimum bias current and get more class A output, but I think this is something you've already considered.
The only areas where I'd think you'd get a better design might be in using a triple EF to reduce the VAS loading and to use better outputs like the mjl4281, reliability should improve as would the ability to drive tougher loads, with possibly fewer output pairs (providing it's stable!). If the heatsinking allowed and was a direction you'd like to move in, you could reduce the value of the RE resistors to increase the optimum bias current and get more class A output, but I think this is something you've already considered.
Dear "5th element",
thank you for your response.
Initially I thought about more class A and the first design steps were based on 0.22 RE resistors. The 0.47 are a tribute to the environment.
Another Idea behind the concept was to design the amplifier so, that it could be driven from half the power supply voltage by nearly the same performance ("green switch"?).
BR, Toni
P.S.: If this design is finished - wether it rocks or not - maybe a 3EF will come. To complete this amplifier we need a power supply, thermal, DC and speaker control and of course last but not least a case. Many hours of DIY will be necessary ...
PSRR testing by LTSpice
Dear keantoken,
THX. Attached a new PSRR test. Hopefully now correct. Are these acceptable values - do we need here better filtering?
BR, Toni
Dear keantoken,
THX. Attached a new PSRR test. Hopefully now correct. Are these acceptable values - do we need here better filtering?
BR, Toni
Attachments
Please don't do this Toni.
Triple EFs are evil. Using them in your excellent design will definitely degrade stability and probably NOT give any better THD. Your design is already better than most designs with triples ... such that it has already attracted favourable comment from triple afficianado, Bob Cordell.
IMHO, you can get even better performance without increasing the attractive simplicity, with different compensation methods. I see you are already doing this with your variations on TPC.
You might like to try evil TMC and ,, if you are very brave, even 'pure Cherry' http://www.diyaudio.com/forums/solid-state/235188-tpc-vs-tmc-vs-pure-cherry.html#post3474671
Could you send me the latest LTspice .ASC for your amp and also the transistor models? I'd like to see if your excellent circuit can be further improved with 'pure Cherry' .. at least in SPICE world
I've PM'd
Last edited:
Please don't do this Toni.
Triple EFs are evil. Using them in your excellent design will definitely degrade stability and probably NOT give any better THD. Your design is already better than most designs with triples ... such that it has already attracted favourable comment from triple afficianado, Bob Cordell.
I don't agree with You. I used triple EF in many amplifiers, without any problem. And the distortion is decreased to 1/10 of the original double EF version.
Especially the triple EF gives lower output impedance before feedback.
Sajti
Please don't do this Toni.
Triple EFs are evil. Using them in your excellent design will definitely degrade stability and probably NOT give any better THD. Your design is already better than most designs with triples ... such that it has already attracted favourable comment from triple afficianado, Bob Cordell.
IMHO, you can get even better performance without increasing the attractive simplicity, with different compensation methods. I see you are already doing this with your variations on TPC.
You might like to try evil TMC and ,, if you are very brave, even 'pure Cherry' http://www.diyaudio.com/forums/solid-state/235188-tpc-vs-tmc-vs-pure-cherry.html#post3474671
Could you send me the latest LTspice .ASC for your amp and also the transistor models? I'd like to see if your excellent circuit can be further improved with 'pure Cherry' .. at least in SPICE world
I've PM'd
Voila!
Note: Cordell-Models.txt must be downloaded from bobs homepage.
TMC: I have tried but got the amp either not stable or with bad performance compared to TPC. Maybe you find the key ...
BR, Toni
Sajti, have you a link to your circuits? Are these distortion figures 'real life' or SPICE?
Toni, thanks for the models.
It is all in the implementation. Most of the output stage distortion is voltage distortion, not drive distortion, so no matter how much you buffer it, miller and Early leakage will conduct its distortion directly into the signal path.
Yes, but the loop gain is increasing, and the higher feedback results lower distortion.
Sajti
You should consult page 8 and the addendum in my paper and the references therein.
Don't forget that here you have only measured +ve rail PSRR. You will need to repeat the test but with the AC source in series with the negative-rail voltage source.
Last edited:
I never said an EF3 wouldn't decrease distortion if properly applied. I'm only pointing out that it takes skill to do this.
Dear Harry,
thanks for hints. Attached a PSRR check also from negative rail - better as from positive rail.
BR, Toni
Attachments
Last edited:
New driver BJT's ...
New driver bjt's (2SC4793/2SA1837 and MJE15032/MJE15033) arrived today! Need matched bjt's.
Will this do with this homebrew "breadboard hellmachine". Can measure 10 pieces of BJT's/JFET's/MOSFET's automatically.
Needs up to 3 USB-controlled multimeters.
New driver bjt's (2SC4793/2SA1837 and MJE15032/MJE15033) arrived today! Need matched bjt's.
Will this do with this homebrew "breadboard hellmachine". Can measure 10 pieces of BJT's/JFET's/MOSFET's automatically.
Needs up to 3 USB-controlled multimeters.
Attachments
Can measure 10 pieces of BJT's/JFET's/MOSFET's automatically.
Needs up to 3 USB-controlled multimeters.
Lol very nice, I like it