I would make R3 alot smaller. Have a look at ASTX's amp. He paralleled the drivers to cope with all that output transistors.
Keep R10 and 12 small and R38 and R39 larger for better distortion figures
Keep R10 and 12 small and R38 and R39 larger for better distortion figures
Last edited:
R13 AND R36.
Also why so much series resistance on the input.
Resistor induced noise can be avoided here
Also why so much series resistance on the input.
Resistor induced noise can be avoided here
R40 is nice and low at 330 ohms, specifically to keep the noise as low as practical,
the values of R2 and R9 completely ruin this by adding 4.3k of Johnson noise and
swamping the efforts of R40, adding by my calcs about 8dB to the noise floor upto
50µVrms across the audio band (ignoring semiconductor contributions).
Thinking again about noise the input devices are probably too noisy - perhaps 2SA992's
are better?
the values of R2 and R9 completely ruin this by adding 4.3k of Johnson noise and
swamping the efforts of R40, adding by my calcs about 8dB to the noise floor upto
50µVrms across the audio band (ignoring semiconductor contributions).
Thinking again about noise the input devices are probably too noisy - perhaps 2SA992's
are better?
Last edited:
-no basestoppers needed Q7/18 , my EF3 DID NOT like that.
-R11 ...0 ? 1K resistor here for a handy CCS measurement.
-No Cascode
-no independant decoupling of the predriver/driver group.
-the "blameless" IPS itself has horrible PSRR on the active (non-CCS) rail.
You could use just 47-100u positive and max out 1K+uf for your negative.
Simple cap multipliers with 1845/992 semi's , 2R's and a little 22uf cap
could give you twice that. I use 100uf/ +TTA/C004's for mine. "Blameless"
still sucks in LF PSSR .... but not as much.
-mje15032/33 can't power 5 pairs !! use some big boys (njw0302/0281).
10 pairs no issue ... EDIT - the best are some of the Sanken fast 100W TO-3p's (as drivers) ,low Cob.
-C10/R31 ??? Simple "blameless" don't need no lag comp.? I have only
used that where I wanted to boost the bandwidth and margin of a symmetric amp out to 3.5 mhZ !!
A multistage IPS like some Sansui's use lead and lag because of the extra
stages add phase delay.
This blameless would be nice at a single Miller (C11 -68pF) for 800khZ UG.
...EF3 DONT need no 11mA VAS. R4 @ 56 , I use 150 (4mA) , I hsve used 3mA- 180R no degradation.
Cold VAS don't matter on a blameless because
most of the circuit cancels out Tc.
My Behringer blameless uses 2 CCS diodes like yours, VAS has a little neg. Tc ... Behringer offset this with the EF Vbe Tc.
2Q CCS is what makes a blameless a blameless ... I use 2 TTA004's for mine.
In fact LTP 2Q CCS is the way to go ( 2 - 2Q CCS's). That's why this
rail is not the PSRR weak point.
...No offset. This amp will not fry speakers .. but will have native offset of up to 50mV. 3 options -
-inject a small mV DC through a trimmer to the DC cap.
-unbalance the local FB of Q4/5 (with a trimmer).
-Servo it , works like the trimmer/DC cap combined.
I opted out , will only servo where I can alter a CCS , or anything else outside the FB route.
...Typical standard Vbe will work with a EF3 , old Sansui EF3 has your vbe.
It works ok until it gets real hot , negative Tc a high temperatures.
A 2Q Vbe , like mine ... 3-4mA 0 - 70C typical.
...C5 -Vbe cap can be just 16-35V ,it is just a 4V shunt with a EF3.
It can be just 1-2.2uF with the VAS running at a reasonable 4-5mA.
OS
Last edited:
phoenix87 When I first asked for help, All the criticism seemed harsh to me. tearing apart the circuit i thought was nice.
The guys bere know their stuff. Make changes and ask more questions, before you know it, you'll have an amp you really proud of.
As for changes, all changes you make are up to you, don't feel pressured into any direction you don't feel like you want to go. Make the critical changes, see which of the improvements you would like to add and go forward with your design.
Each amp can be different in which features it includes. I'm guessing that an amp that could drive 4R at 400W reliably and with good sound is the main goal.
Regards
The guys bere know their stuff. Make changes and ask more questions, before you know it, you'll have an amp you really proud of.
As for changes, all changes you make are up to you, don't feel pressured into any direction you don't feel like you want to go. Make the critical changes, see which of the improvements you would like to add and go forward with your design.
Each amp can be different in which features it includes. I'm guessing that an amp that could drive 4R at 400W reliably and with good sound is the main goal.
Regards
Last edited:
It is depend of your priority:
1. Slew rate
2. Good harmonic profile
3. THD
4. PSRR
5. Simplicity
etc.
In this forum at least 2 good design of Blameless amplifier by Ostripper and Astx. You can read those thread.
1. Slew rate
2. Good harmonic profile
3. THD
4. PSRR
5. Simplicity
etc.
In this forum at least 2 good design of Blameless amplifier by Ostripper and Astx. You can read those thread.
You should check the clipping behavior. At high powers it isn't as much of an issue because you won't clip, but when I was experimenting with some blameless designs, the clipping behavior was a problem on one rail. I never found a good solution and wouldn't want to invest a lot on a flawed design.
Blameless clipping ... no issue.
(below) VAS I at clip. No SOA violations here.
TOP is CCS I , middle is VAS , bottom is output.
(below 2) larger R17 for VAS beta driver Q12. >2.7k is where slew rate
declines but device collector I has a limit.
D3 - BAV21 is low capacitance , this clips nearly as well as a symmetric.
Clipped the real world design 24hrs 100W , CCS/VAS was cold. Q12 got
just warm. Q10/11 got cold (room temp.)
PS -with BAV21 8PPM without 5PPM .... worth it.
edit - independant LTP and VAS CCS's are worth it , too. Design in this thread,
clipping would affect the LTP CCS.
The real world test ... 3 pair MT-200 Blameless at clip ... lighting up a toaster oven to incandesance
for 24 hours ... had toast.
OS
(below) VAS I at clip. No SOA violations here.
TOP is CCS I , middle is VAS , bottom is output.
(below 2) larger R17 for VAS beta driver Q12. >2.7k is where slew rate
declines but device collector I has a limit.
D3 - BAV21 is low capacitance , this clips nearly as well as a symmetric.
Clipped the real world design 24hrs 100W , CCS/VAS was cold. Q12 got
just warm. Q10/11 got cold (room temp.)
PS -with BAV21 8PPM without 5PPM .... worth it.
edit - independant LTP and VAS CCS's are worth it , too. Design in this thread,
clipping would affect the LTP CCS.
The real world test ... 3 pair MT-200 Blameless at clip ... lighting up a toaster oven to incandesance
for 24 hours ... had toast.
OS
Attachments
Last edited: