a wonderfull sziklai - Page 3

megajocke · 20th December 2008, 10:05 AM

Ooh, but that's not too scary compared to what I've seen on some imported cheap amps. QSC RMX1450 uses 4 pairs on +-78V and can drive 2 ohm loads (but the rails sag a lot in 2 ohms) so your amp doesn't seem extremely underdesigned at least. Also, Crest CA-4 uses 4 pairs of it's predecessors 2SC3281/2SA1302 on just above +-80V IIRC. 90V is probably pushing it though, second breakdown limit degrades pretty fast with increasing voltage!

The really bad one I've seen has 4 pairs on +-120V

Actually I don't know if any amount of pairs of these is a good solution at +-120V due to second breakdown. Series connected transistors or rail switching would help a lot... The amp also has emitter resistors that are too small. The output stage doesn't have any bias normally but if the transistors were ever to heat up enough to start passing bias current it's instant thermal runaway and destruction!

Thought about using MJL2119x in that amp? I've had trouble finding them at reasonable prices though.

m2003br · 20th December 2008, 12:42 PM

Megajocke, with Toshibas of the same batch, you have a very good matching, no need for selection. very unlike that oldies american TO-3.
About sakis circuit, you already have a good compensation of current sharing thanks to separated EMITTER (yes, in a sziklai, colectors turn to be emitters, because the master is the driver transistor) resistors combined with individual feedback resistors of 4.7R.
Mr. Douglas Self, in the book, shows a sziklai output amp (the "Load Invariant") with 2 directly paralelled transistors, yes, without any resistors!
Marcos

fotios · 20th December 2008, 12:43 PM

Megajoke

Your remarks are identical to mine. It seems you have a lot of experience about PA amplifiers... i am in this game about 25 years... many times with bitter experience from which i got the real knoweledge. Accordingly i have the sense, that it is not wise the supply level to exceeds the +/-90Vdc... for the given semiconductors on the market today. The only possibility offered to break this limit, it is the cascade.
From the other side, considering - let us to say - that we have unlimited paralleled transistors so the output can sustain any current, +/-90V supply translated in: 510Wrms/8Ù or 1020Wrms/4Ù.
Is there any speaker which needs a so high REAL power level to give its maximum SPL? Unless it has a bad designed box and crossover, which can consume the 50% of power. I have seen many times such type speakers, the most of was Hi-Fi maden from irrelevant amateurs. These guys, they confuse usually the maximum SPL obtained by a speaker regardless of the power needed.
I remember before 20 years, two guys, they sold 8 diy speakers equiped with Vifa units in a club, and i sold a Peavey CS-1000X (according to their demands in power

) and the result was broken terminal wires in woofers and burned midranges & tweeters when the people exceeded the 150 persons... of course, guilty considered the CS-1000X for the destruction

... you know from such cases.

Regs
Fotios

megajocke · 20th December 2008, 01:46 PM

Quote:

Originally posted by m2003br
Megajocke, with Toshibas of the same batch, you have a very good matching, no need for selection. very unlike that oldies american TO-3.
About sakis circuit, you already have a good compensation of current sharing thanks to separated EMITTER (yes, in a sziklai, colectors turn to be emitters, because the master is the driver transistor) resistors combined with individual feedback resistors of 4.7R.
Mr. Douglas Self, in the book, shows a sziklai output amp (the "Load Invariant") with 2 directly paralelled transistors, yes, without any resistors!
Marcos

I agree that matching should not be needed as you say. Seen from the outside of the sziklai you are kind of correct but inside it those are still collector resistors.

Vbe is the same for both output devices - if matched well that means same collector current will flow through both. But it is very sensitive to temperature variations and bias current. The collector resistors shown do nothing to help current share. Unless the transistor is working in its saturation region Vce chagnes won't affect the collector current much. Only base-emitter voltage and temperature matters.

The higher collector current the higher transconductance the transistors get, which means more positive thermal feedback the higher the collector current gets. At 100mA collector current the transconductance will be 4S. (Ic/Vt) This will be stable. But at 1A collector current the transconductance will be almost ten times higher giving lots of positive thermal feedback, shifting all current to one of the transistors which then will take all load because it is much hotter and has the same vbe as its parallell transistor which might not even turn on.

Adding emitter resistors of 0.22 ohms will make it impossible for transconductance to exceed ~4S, making it stable.

megajocke · 20th December 2008, 01:55 PM

Quote:

Originally posted by fotios
Megajoke

Your remarks are identical to mine. It seems you have a lot of experience about PA amplifiers... i am in this game about 25 years... many times with bitter experience from which i got the real knoweledge. Accordingly i have the sense, that it is not wise the supply level to exceeds the +/-90Vdc... for the given semiconductors on the market today. The only possibility offered to break this limit, it is the cascade.
From the other side, considering - let us to say - that we have unlimited paralleled transistors so the output can sustain any current, +/-90V supply translated in: 510Wrms/8Ù or 1020Wrms/4Ù.
Is there any speaker which needs a so high REAL power level to give its maximum SPL? Unless it has a bad designed box and crossover, which can consume the 50% of power. I have seen many times such type speakers, the most of was Hi-Fi maden from irrelevant amateurs. These guys, they confuse usually the maximum SPL obtained by a speaker regardless of the power needed.
I remember before 20 years, two guys, they sold 8 diy speakers equiped with Vifa units in a club, and i sold a Peavey CS-1000X (according to their demands in power ) and the result was broken terminal wires in woofers and burned midranges & tweeters when the people exceeded the 150 persons... of course, guilty considered the CS-1000X for the destruction ... you know from such cases.

Regs
Fotios

Drivers made for Hifi in club speakers, with a big amplifier, and being driven by a DJ doesn't sound like a good combination

But as long as average power isn't too high (= not a DJ trying to keep the clip lights constantly on) those high powered amplifiers can be useful as speakers can take much more peak power than average power as long as duty cycle is low...

catalin · 20th December 2008, 04:51 PM

Hello !
I'm smiling and I'm glad when i have seen this post.I do not know how my design reached an greek forum.The original thread is http://www.elforum.ro/viewtopic.php?...4101&start=60, Romanian
forum.This amp is a compromise between the best sound (slew
rate, THD, phase, linearity), power consumed (it sounds better than a
amp in Class A) and the number of transistors.I wish to make this amp in mirror so it will increase the slew rate and the cross over distortion will be slower.
Your ideas are welcome .
I put resistance for the collector because I want to
implement protection in a final stage.if you don't want a protection you can put the resistor in emitor ..
Why do you want to bump 2n5401?
I attached the last version wich is more stable ,and it has more dynamics.The slew rate measured with one pair of 2sc5200/1943 was 70V/us .The thd is 0.001%.The second harmonic is almost egual with the third ..You can look at the romanian forum or if you want I can post the other measurements and results...
I named this amp JOY200 ...
Have a nice day !

catalin · 20th December 2008, 04:51 PM

i forgot to tell.I have test this amp with 2 speaker(8ohms) in parallel(4ohms) at full power(200w)

megajocke · 20th December 2008, 09:00 PM

Did you check load sharing between parallell output transistors?

A good test would be expected lowest impedance dummy load, 1/2 output power (about 2/3 clipping voltage) sine wave. Measure voltage drop over collector resistors with average reading voltmeter during this test.

A cruder test would be comparing the case temperature of the parallelled power transistors when playing very loud music.

The collector resistors won't help load sharing like emitter resistors do. Even if it doesn't share it will work for a while, but don't expect it to be reliable at high power.

sakis · 21st December 2008, 10:07 AM

Quote:

Originally posted by catalin
Hello !
I'm smiling and I'm glad when i have seen this post.I do not know how my design reached an greek forum.The original thread is http://www.elforum.ro/viewtopic.php?...4101&start=60, Romanian
forum.This amp is a compromise between the best sound (slew
rate, THD, phase, linearity), power consumed (it sounds better than a
amp in Class A) and the number of transistors.I wish to make this amp in mirror so it will increase the slew rate and the cross over distortion will be slower.
Your ideas are welcome .
I put resistance for the collector because I want to
implement protection in a final stage.if you don't want a protection you can put the resistor in emitor ..
Why do you want to bump 2n5401?
I attached the last version wich is more stable ,and it has more dynamics.The slew rate measured with one pair of 2sc5200/1943 was 70V/us .The thd is 0.001%.The second harmonic is almost egual with the third ..You can look at the romanian forum or if you want I can post the other measurements and results...
I named this amp JOY200 ...
Have a nice day !

it is very nice to see you here .... and i apologize since i didnt know that this is your design to ask you first before posting ....
it is very nice to have you here and something is telling me that this will be a wonderfull issue to talk about ....

about SOA and fotios previous coments versus andrew t coments ....

at this point i will go with fotios .... may be andrew as usual have a detailed figure of soa of any amp ( what i like with andrew is the he never drop a comment without supporting it with data ) i find this very usefull .....

to get somewhere i mean that we have to take in mind that al these chinese amps and also crown , qsc, beringer and clones are on the street and WORKING !!!!!! so that means that andrew's estimation about SOA is correct but OPTIMUM a litle bit less prooved that didnt harm anyone ....

lets see the amp improved by the aythor now ....

thanks people

jacks76 · 22nd December 2008, 03:10 AM

Big Power Amplifier is always interesting, regardless that it's really need on the real world.
When DC Supply high voltage can reached more and more (economically) and Higher Voltage PSU Capacitor more Available Nowadays (still expensive).
Then SOA is next Problem, that in fact ,Research & development in transistor world (Higher Voltage , Ampere etc) seems stack in Place. - not to say - already reached the limit.

Next Heatsink, Wiring etc, Problem seem depend on US rather then Component itself.

we always offered to use more Amplifier unit on PA Duty from the producers (factory) , rather than Bigger output amplifier to handle huge amount of speakers.

As a DIY'ers , for ourself , then for. a little bit Commercial purpose,
I always faced to multiply BJT Output Transistor, as a part of the high demand on Bigger Power Amp in the fact.

so , as a DIY'ers isn't it challenge us ? or
we stay still until next generation Of Transistor or Electronic ?
Of Couse that Quality aproach must on the same way to go with the 'Quantity ' demand.

Really sorry , if it's look like amateurs approach or less knowledge way of view.

20th December 2008, 12:42 PM	#22
m2003br diyAudio Member Join Date: Dec 2004 Location: Sao Paulo	current sharing Megajocke, with Toshibas of the same batch, you have a very good matching, no need for selection. very unlike that oldies american TO-3. About sakis circuit, you already have a good compensation of current sharing thanks to separated EMITTER (yes, in a sziklai, colectors turn to be emitters, because the master is the driver transistor) resistors combined with individual feedback resistors of 4.7R. Mr. Douglas Self, in the book, shows a sziklai output amp (the "Load Invariant") with 2 directly paralelled transistors, yes, without any resistors! Marcos

Thread Tools	Search this Thread
Show Printable Version Email this Page	Search this Thread: Advanced Search

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Sziklai output	Moby	Solid State	68	27th January 2011 10:39 AM
Hybrid Sziklai CFP	Stee	Solid State	0	27th December 2008 10:53 AM
ccs-loaded sziklai	ionomolo	Solid State	132	30th June 2008 06:39 AM
My DIY project - Wonderfull Loudspeakers !	Emeric	Multi-Way	6	9th May 2008 02:38 AM

New To Site?	Need Help?
Register to Participate Search Privacy Statement Contact Us	Frequently Asked Questions Did you forget your password? Mark Forums Read


	Please consider donating to help us continue to serve you. Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving

20th December 2008, 10:05 AM	#21
megajocke diyAudio Member Join Date: Jan 2003	Ooh, but that's not too scary compared to what I've seen on some imported cheap amps. QSC RMX1450 uses 4 pairs on +-78V and can drive 2 ohm loads (but the rails sag a lot in 2 ohms) so your amp doesn't seem extremely underdesigned at least. Also, Crest CA-4 uses 4 pairs of it's predecessors 2SC3281/2SA1302 on just above +-80V IIRC. 90V is probably pushing it though, second breakdown limit degrades pretty fast with increasing voltage! The really bad one I've seen has 4 pairs on +-120V Actually I don't know if any amount of pairs of these is a good solution at +-120V due to second breakdown. Series connected transistors or rail switching would help a lot... The amp also has emitter resistors that are too small. The output stage doesn't have any bias normally but if the transistors were ever to heat up enough to start passing bias current it's instant thermal runaway and destruction! Thought about using MJL2119x in that amp? I've had trouble finding them at reasonable prices though.

20th December 2008, 12:43 PM	#23
fotios diyAudio Member Join Date: Feb 2007 Location: ΔΡΑΜΑ - North Greece	Megajoke Your remarks are identical to mine. It seems you have a lot of experience about PA amplifiers... i am in this game about 25 years... many times with bitter experience from which i got the real knoweledge. Accordingly i have the sense, that it is not wise the supply level to exceeds the +/-90Vdc... for the given semiconductors on the market today. The only possibility offered to break this limit, it is the cascade. From the other side, considering - let us to say - that we have unlimited paralleled transistors so the output can sustain any current, +/-90V supply translated in: 510Wrms/8Ù or 1020Wrms/4Ù. Is there any speaker which needs a so high REAL power level to give its maximum SPL? Unless it has a bad designed box and crossover, which can consume the 50% of power. I have seen many times such type speakers, the most of was Hi-Fi maden from irrelevant amateurs. These guys, they confuse usually the maximum SPL obtained by a speaker regardless of the power needed. I remember before 20 years, two guys, they sold 8 diy speakers equiped with Vifa units in a club, and i sold a Peavey CS-1000X (according to their demands in power ) and the result was broken terminal wires in woofers and burned midranges & tweeters when the people exceeded the 150 persons... of course, guilty considered the CS-1000X for the destruction ... you know from such cases. Regs Fotios

20th December 2008, 04:51 PM	#27
catalin diyAudio Member Join Date: Jun 2007	i forgot to tell.I have test this amp with 2 speaker(8ohms) in parallel(4ohms) at full power(200w)

20th December 2008, 09:00 PM	#28
megajocke diyAudio Member Join Date: Jan 2003	Did you check load sharing between parallell output transistors? A good test would be expected lowest impedance dummy load, 1/2 output power (about 2/3 clipping voltage) sine wave. Measure voltage drop over collector resistors with average reading voltmeter during this test. A cruder test would be comparing the case temperature of the parallelled power transistors when playing very loud music. The collector resistors won't help load sharing like emitter resistors do. Even if it doesn't share it will work for a while, but don't expect it to be reliable at high power.

22nd December 2008, 03:10 AM	#30
jacks76 diyAudio Member Join Date: Sep 2006	Big Power Amplifier is always interesting, regardless that it's really need on the real world. When DC Supply high voltage can reached more and more (economically) and Higher Voltage PSU Capacitor more Available Nowadays (still expensive). Then SOA is next Problem, that in fact ,Research & development in transistor world (Higher Voltage , Ampere etc) seems stack in Place. - not to say - already reached the limit. Next Heatsink, Wiring etc, Problem seem depend on US rather then Component itself. we always offered to use more Amplifier unit on PA Duty from the producers (factory) , rather than Bigger output amplifier to handle huge amount of speakers. As a DIY'ers , for ourself , then for. a little bit Commercial purpose, I always faced to multiply BJT Output Transistor, as a part of the high demand on Bigger Power Amp in the fact. so , as a DIY'ers isn't it challenge us ? or we stay still until next generation Of Transistor or Electronic ? Of Couse that Quality aproach must on the same way to go with the 'Quantity ' demand. Really sorry , if it's look like amateurs approach or less knowledge way of view.