I have not been able to blow it up yet..
I had my earlier version on 60v rails , and overdrove that one to collapse the rails , no blown components. we could baker clamp it , but I don't see many other DIY amps with those kind of considerations (quasi's , dx, even the symasym). The back to back diodes are all right , and protect against gross overdrive. No burning amps here !! OS
Signal Ground is part of the input flow and return circuit.Rafael.luc said:
The Ground can be exposed either at the amplifier or at the far end of the interconnect.
This exposed part must be connected to the Safety Earth.
The diodes should provide the high current route from Signal Ground to Safety Earth.
Signal diodes are the wrong choice for this duty.
1n4001 would be better.
1n5401 better still.
Do we need the 22r?
What effect does it have on the amplifier's performance?
PCB my attempt
Hi ,
I'am ready, to hear some comments, regarding my PCB for supersym , and
I will fix any error http://i39.tinypic.com/jgirk9.jpg
Regards Alex mm.
Hi ,
I'am ready, to hear some comments, regarding my PCB for supersym , and
I will fix any error http://i39.tinypic.com/jgirk9.jpg
Regards Alex mm.
AndrewT said:
I do not understand why anti-parallel two diodes, do to improve Safety Earth.
AndrewT said:
Discussion started here:
http://www.diyaudio.com/forums/showthread.php?postid=739086#post739086
But, I do not agree with the use...
Hi ostripper,
Let me know, it would be great to get drivers with high beta.
Hi Piersma,
-Chris
I think we might be talking about different parts (I hope). The beta range you are reporting is for the differential transistors, right? I was talking about the cascode transistors. I haven't looked them up, but a high voltage TO-220 normally does not have a high beta.
Let me know, it would be great to get drivers with high beta.
Hi Piersma,
No need. A touch of thermal compound and a little heat shrink tubing and you have created a "micro climate".
-Chris
Have you tried clicking on it??... it links to the full size one on photobucket ..(no popups)
That is beautiful , man I might copy some things from it..
Hummm , I guess we are not (talking the same parts) The cascode's are to-92(not to-220) 2sc - KSC 1845 (120v Vce) they are in 4 gain groups ... P to U - 200 to 600 Hfe groups... http://www.fairchildsemi.com/ds/KS/KSC1845.pdf
The Cascode just passes 1.8mA per side and has a voltage drop of 70- 6.2v (63.8v Vce) why would you need 220's ?
The only to-220's I can see are the mje15033/32's.
I just have a dab of compound with 3/8" Hs tubing , they are "front to back" but close together. On first turn on, before trimming with just DMM matching , I had just 3-5mv offset on either channel. after 5 weeks of heavy use , 1-2mv offset in any environment.
OS
I must add this comment to any constructors.If you go with alex's good looking board or the one I have not finished yet, and you want a tested real life amp , the cascode/JFET has NOT been built yet.
The base amp , WITHOUT the additions is up and working. While I am 99% sure the cascode will work and still be stable , the BASE design is 100% (listening to it now ).
I noticed the cascode and VAS was what determined OLG , UG , phase margin. Replacing the differential with anything (MOSFET , JFET, more KSC's) had very little effect on performance.
One might want to try slight variances on component selection. So far I have tried 47 and 68pF cdoms.
68,82 and 100R for R16 (main degen for the VAS collectors).
with and without C21 (10p).
ALL these affect the character of the sound while maintaining great stability. I also have experimented with the OPS bias (60-70mA per device sounds the best).
If doing the JFET/ cascode , the wise man builds the whole amp EXCEPT for the OPS. Just solder/tack 2 33R resistors from the MJE emitters to the output rail with the MJE's on heatsinks. With a 330R dropping resistor , one can listen to the amp with headphones. If you only read 1V emitter to emitter on the drivers , put in the OP devices and be happily underbiased. No burning amps..
OS
The base amp , WITHOUT the additions is up and working. While I am 99% sure the cascode will work and still be stable , the BASE design is 100% (listening to it now
).I noticed the cascode and VAS was what determined OLG , UG , phase margin. Replacing the differential with anything (MOSFET , JFET, more KSC's) had very little effect on performance.
One might want to try slight variances on component selection. So far I have tried 47 and 68pF cdoms.
68,82 and 100R for R16 (main degen for the VAS collectors).
with and without C21 (10p).
ALL these affect the character of the sound while maintaining great stability. I also have experimented with the OPS bias (60-70mA per device sounds the best).
If doing the JFET/ cascode , the wise man builds the whole amp EXCEPT for the OPS. Just solder/tack 2 33R resistors from the MJE emitters to the output rail with the MJE's on heatsinks. With a 330R dropping resistor , one can listen to the amp with headphones. If you only read 1V emitter to emitter on the drivers , put in the OP devices and be happily underbiased. No
burning amps.. OS
Where to start...
1. output is taken off one side , as is zoble , as is NFB takeoff.
(hint), treat the output as you do the star ground.
2. Main star ground is close , but has to be TOTALLY symmetrical . To test this I attached my speaker ground just 1" to one side of the star and heard a little hum. The rail cap pulses cancel out EXACTLY half way between the two rail caps.
3. This is a little bit my fault .. BIAS adjust trimmer is a 1K unit (565 = 70ma per device.)
4 . we might want to add some reverse biased "flyback" protection diodes from rails to output bus.
Good things ..
1. I like all the capacitance on board , but having the AC and power rectifiers there too is over integrated. The better way is 35A heatsinked offboard bridges.
2 . No real circuit errors .. it would work , but you might have higher THD because of error 1.
3.It looks pretty compact , but will it fit flat on a 5" heatsink??
OS
I had forgotten that discussion.Rafael.luc said:
I agree that the Signal Ground can benefit from some form of Audio isolation from the Power Ground.
I insist that the Signal Ground (if exposed at any location) must be connected to Safety Earth.
What I am asking is:
Can we just use a pair of inverse parallel diodes to connect the input circuit to the Safety Earth?
Does the circuit/amplifier NEED the resistor from Signal Ground to Power Ground?
If an unsafe voltage accidentally contacts the Signal Ground somewhere along it's route then the diodes conduct, with either polarity, to Safety Earth when the voltage >~600mV
Without the resistor a potential difference between power and signal ground of up to ~600 mV can build up. That would offset the correction signal and affect the NFB function. The lower the signal voltage, the worse the effect.
E. g. with the feedback combination 27k/1k shown in the last schematic 60 V output signal should give 2,14 V correction signal, but can give anything between 2,14 and 2,74 (up to 28 % difference). 6 V output signal should give 0,214 V correction signal, but can give anything between 0,214 and 0,814 (up to 280 % difference). 0,6 V output signal should give 0,0214 V correction signal, but can give anything between 0,0214 and 0,6214 (up to 2800 % difference).
E. g. with the feedback combination 27k/1k shown in the last schematic 60 V output signal should give 2,14 V correction signal, but can give anything between 2,14 and 2,74 (up to 28 % difference). 6 V output signal should give 0,214 V correction signal, but can give anything between 0,214 and 0,814 (up to 280 % difference). 0,6 V output signal should give 0,0214 V correction signal, but can give anything between 0,0214 and 0,6214 (up to 2800 % difference).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.