I can see the argument for "1st watt" ideas, as that is where most listening will take place. Thermal stability could be a problem, although CFP might be easier than a Darlington in this respect. However, if you push a CFP up into the high current region then it is really no different from a Darlington. The gm boost a CFP has is only there in the low-intermediate current region. I guess a CFP with too much current will be better than a Darlington with not enough.
This is a commonly overlooked ADVANTAGE of CFP. Having to deal only with the smaller thermal time constants of the drivers makes the bias regulation easier and possibly more precise.
This lets you work with smaller bias currents, just over the limit of cross-over distorsion, with very good distorsion figures nevertheless. I believe the old way of using a generator and a scope raising the bias slowly until cross over disappears (with patience and waiting for temperature to setup) is still the best one, at least for DIY.
If I want a little perfect stove then I´ll happily apply for one of the fantastic class A of Mr Pass.
How many ugly!
Now your point is a little mor clear to me. I beg your pardon, but I am old and not so fast.
Now I understand that you just do not like BJTs and like only (lateral) FETs. I am sure many will agree with you and many will disagree. I am not in favour of either. But do not worry, this thread is about a Sziklai BJT amplifier, and that will not change, that you like or not.
And I DO NOT want to turn this tread in a a sterile dispute BJT VS FET.
If you want to do so , I beg you to open your own thread on the subjet and let me alone. I will happily partecipate in your thread speaking about majority carriers, and why not, 1/f noise, gm (or lack of thereof) and so on. Thanks for your understanding.
Last edited:
Yes, the choice here is to get the best possible with the smaller number of active components. So, of course, you have to search for good ones.
And as a VAS, those are hard to beat! Against the old good BD139/40 just on top of my mind they have higher voltage, smaller Cob, higher Ft.
About availability, Farnell has them, I beleive, althogh arent´cheap as BD140s, but you cannot have everything....
Finally I haven´t simulated the thing, nor I plan to do it (I do not know if it makes sense for so simple circuit) , but if somebody has the time and skills to do he´s welcome!
Layout
Here is a components layout. PCB is to be double sided, metallized holes, but can be easily reverted to single side for domestic etching.
I am still refining it though.
Some rough ampli characteristics:
RMS output power, with 2 x TO3 pairs:
150W on 8 Ohm (50V rails)
200W on 4 Ohm (42 V rails)
RMS Output power with single pair:
80 W on 8 Ohm (42 V rails)
100 W on 4 Ohm (35 V rails)
Closed loop gain : 22
Open loop gain 500-1000
THD , BP, etc. to be measured on prototype.
Here is a components layout. PCB is to be double sided, metallized holes, but can be easily reverted to single side for domestic etching.
I am still refining it though.
Some rough ampli characteristics:
RMS output power, with 2 x TO3 pairs:
150W on 8 Ohm (50V rails)
200W on 4 Ohm (42 V rails)
RMS Output power with single pair:
80 W on 8 Ohm (42 V rails)
100 W on 4 Ohm (35 V rails)
Closed loop gain : 22
Open loop gain 500-1000
THD , BP, etc. to be measured on prototype.
PCB
Here is the PCB
It fits half Europa size (80 x100) mm with some allowance for de-panelization, so most PCB factories can do 2 o those boards in a single order.
Making it single side requires only to make a U-shaped GND trace atounf the board abd re-route the two missing tranistors tracks to the left of the boatd (or make a couple of bridges. Same for the Zobel capacitor ground, if fitted on the PCB (usually I fit it directly on the posts).
Here is the PCB
It fits half Europa size (80 x100) mm with some allowance for de-panelization, so most PCB factories can do 2 o those boards in a single order.
Making it single side requires only to make a U-shaped GND trace atounf the board abd re-route the two missing tranistors tracks to the left of the boatd (or make a couple of bridges. Same for the Zobel capacitor ground, if fitted on the PCB (usually I fit it directly on the posts).
Great-I follow it all--The emitter resistors were hiding under the drivers, in my quick look.
(To me an emitter resistor is what it is regardless of being connected to a collector)
I do not follow the logic of using a high gain/ high current trans on the vbe... its job after all, is to provide ~1.25 v of voltage drop, interbase. not much current either--? Mpsa 06 fine? Nor do I understand the cap people always use here,(220n) but there must be a reason? In the push-pull of diy-audio, it's nice to find a defense of 'ol SZ, the compound man.
(To me an emitter resistor is what it is regardless of being connected to a collector)
I do not follow the logic of using a high gain/ high current trans on the vbe... its job after all, is to provide ~1.25 v of voltage drop, interbase. not much current either--? Mpsa 06 fine? Nor do I understand the cap people always use here,(220n) but there must be a reason? In the push-pull of diy-audio, it's nice to find a defense of 'ol SZ, the compound man.
looked at circuit.. sehr gut !
Only a couple things .... 2sa970/c2240 are "end of life" devices. All 2sc1845's would do the same thing. I did not know what you were using the blue led for ? a single red with Q4 - 220Re would do the same job as a CCS reference. since it is not a current sourced VAS the input stage current is crucial , especially using a Sziklai. Either double diodes or the red led will set up the negative coefficient for the input stage. A 2Q CCS at input would keep the VAS current even more constant. On both my BX (bootstrapped) and EX (balanced) just .02 ma input CCS current change will give .1ma VAS change. On the AX (blameless) , one could adjust the input CCS from 1ma to 4 ma total with only <.005ma change in the VAS.Decouple Q4 as well ,earth to 10k / 10uf to neg rail/ 15K to led will get you another 6-10db PSRR ... just 2 more parts.
AND the BIG one ... TMC will not work on this amp. It has insufficient open loop gain. (no CM and no beta enhanced VAS - gain of 50-60 at most.) with a non beta enhanced single 2sa1381 (F grade/ 300Hfe)the best one could get is 70db OLG with low input pair degeneration. TMC needs 90+ to be really effective (AX has 110db+ OLG ! ).
OS
Only a couple things .... 2sa970/c2240 are "end of life" devices. All 2sc1845's would do the same thing. I did not know what you were using the blue led for ? a single red with Q4 - 220Re would do the same job as a CCS reference. since it is not a current sourced VAS the input stage current is crucial , especially using a Sziklai. Either double diodes or the red led will set up the negative coefficient for the input stage. A 2Q CCS at input would keep the VAS current even more constant. On both my BX (bootstrapped) and EX (balanced) just .02 ma input CCS current change will give .1ma VAS change. On the AX (blameless) , one could adjust the input CCS from 1ma to 4 ma total with only <.005ma change in the VAS.Decouple Q4 as well ,earth to 10k / 10uf to neg rail/ 15K to led will get you another 6-10db PSRR ... just 2 more parts.
AND the BIG one ... TMC will not work on this amp. It has insufficient open loop gain. (no CM and no beta enhanced VAS - gain of 50-60 at most.) with a non beta enhanced single 2sa1381 (F grade/ 300Hfe)the best one could get is 70db OLG with low input pair degeneration. TMC needs 90+ to be really effective (AX has 110db+ OLG ! ).
OS
Last edited:
Well, ok for the Re .
Then, for the Vbe I just used one of those (too) good TO220 devices because
1) It is easy to thermally couple it to the same H/S.
2)It has the same B-E junction of at least one of the drivers, so it would follow better the Vbe changes. You are right, probably a BD 139/140 (I do not have MPSA handy here) would do just as fine.
About the cap, I also did not put it on in a first draw. But then I have seen in the Bob Cordell's book that, in effect, the Vbe circuit performs bad at high frequencies, so it needs to be shunted with a capacitor. I was always suspicious of that, and he has given me a good explanation. So, capacitor added.
It is nice to see that there are still people that appreciate the old good Sziklai, a simple way to obtain well performing OP stages.
Thank you for your comments !
Danke!
Then about the 2sc2240 it just happens that I have a bunch of them carefully selected, they are very good! BTW you still can order them from a local big dealer (Distrelec) for cheap. I have no experience with the 1845, I am sure you are right. For the CCS I used the 2N5551 because it is more robust power-wise , and I want to experiment with higher currents for the LTP. I agree that a 2Q would be the best, but I wanted to keep it simple.
Mabe a darligton? I have around some good old BDX53C (hfe 2500) but it seems an overkill. The decoupling capacitor is probably a good idea, especially if it can be done with a small one.
About the blue LED, I am still not convinced that the RED is better. With the BLUE (or the white, even more) you can use a much bigger Re, that helps a lot output impedance . I won't think that thermal beaviour is the issue, but maybe I am wrong. Also note that the LTP transistors are intended to be strictly MATCHED and that the Re trimmer is just for fine-tuning. Maybe also the input transistor bias network has to be tweaked for best results. As I said this is a circuit intended for DIY, have fun!
Then the BIG one.
Yes, I do not espect more than 60 dB OLG in this configuration.
I want to experiment with higher VAS / LTP currents, lesser degeneration, but I know that with higher gains the stability issues would grow.
And I am already hearing the hurdle that is already singing - we told youuuu, CFP is unstable by definitiooooon. I want to prove them wrong and that is why I am bringing this circuit a little on its limits. But I am convinced that if I can make it work (and I WILL) it has to sound GREAT!
So about TMC what do you espect that will be the troubles that I am going to face with lesser gain?
Thank you so much
WOW, OS. This is not a comment, but a complete review, I really appreciate.
Then about the 2sc2240 it just happens that I have a bunch of them carefully selected, they are very good! BTW you still can order them from a local big dealer (Distrelec) for cheap. I have no experience with the 1845, I am sure you are right. For the CCS I used the 2N5551 because it is more robust power-wise , and I want to experiment with higher currents for the LTP. I agree that a 2Q would be the best, but I wanted to keep it simple.
Mabe a darligton? I have around some good old BDX53C (hfe 2500) but it seems an overkill. The decoupling capacitor is probably a good idea, especially if it can be done with a small one.
About the blue LED, I am still not convinced that the RED is better. With the BLUE (or the white, even more) you can use a much bigger Re, that helps a lot output impedance . I won't think that thermal beaviour is the issue, but maybe I am wrong. Also note that the LTP transistors are intended to be strictly MATCHED and that the Re trimmer is just for fine-tuning. Maybe also the input transistor bias network has to be tweaked for best results. As I said this is a circuit intended for DIY, have fun!
Then the BIG one.
Yes, I do not espect more than 60 dB OLG in this configuration.
I want to experiment with higher VAS / LTP currents, lesser degeneration, but I know that with higher gains the stability issues would grow.
And I am already hearing the hurdle that is already singing - we told youuuu, CFP is unstable by definitiooooon. I want to prove them wrong and that is why I am bringing this circuit a little on its limits. But I am convinced that if I can make it work (and I WILL) it has to sound GREAT!
So about TMC what do you espect that will be the troubles that I am going to face with lesser gain?
Thank you so much
At 60-70db gain , there would be no difference THD wise between TMC or conventional miller compensation. TMC "trades off" extra gain (you need 90+ to have "extra") for reduced HF THD. So , at lower gains ... it is a useless endeavor. You would have no troubles , amp would respond the same as the 2 miller caps in series (conventional miller comp.)
OS
Ok, it is more or less what I had tought. I want to make some measurements on the BP (gain) to see if it is somehow influenced by this second pole. I mean it might straigten up a little bit. We'll see. Thanks for the nice infos.
what about standard TPC?
One moment, now OS make me to think to another way.
If I remeber well old standard TPC works fine with lower gains.
I mean with a VAS gain of 50-100. For the circuit it should be enough to disconnect the resistor from the OP and connect it to the (clean) GND.
Maybe also adding a static gain resistor would help, something like a 100K. I do not want to restart a discussion that has been exaustively done in another thread but ist this one of the few cases when TPC works better than TMC? Please be specific if you dare to answer.
.. thank you
One moment, now OS make me to think to another way.
If I remeber well old standard TPC works fine with lower gains.
I mean with a VAS gain of 50-100. For the circuit it should be enough to disconnect the resistor from the OP and connect it to the (clean) GND.
Maybe also adding a static gain resistor would help, something like a 100K. I do not want to restart a discussion that has been exaustively done in another thread but ist this one of the few cases when TPC works better than TMC? Please be specific if you dare to answer.
.. thank you
Start reading Cordell's thread here : http://www.diyaudio.com/forums/solid-state/171159-bob-cordells-power-amplifier-book-18.html
Required reading.....
My AX has the CMC , TPC , TMC "universality" (below 1-look at the board bottom) . I know this is subjective , but guess what I am listening to now ? Without the beta enhancement or mirror , I suppose TPC is the only enhanced compensation usable. Make a board and leave it "open ended" to allow for CMC or TPC.
OS
Required reading.....
My AX has the CMC , TPC , TMC "universality" (below 1-look at the board bottom) . I know this is subjective , but guess what I am listening to now ?
Without the beta enhancement or mirror , I suppose TPC is the only enhanced compensation usable. Make a board and leave it "open ended" to allow for CMC or TPC.OS
Attachments
Yeah, Bob Cordell wrote an awesome book. I had downloaded the sample on my Kindle and it looks to me the best one on Power Amps.Start reading Cordell's thread here : http://www.diyaudio.com/forums/solid-state/171159-bob-cordells-power-amplifier-book-18.html
Required reading.....
My AX has the CMC , TPC , TMC "universality" (below 1-look at the board bottom) . I know this is subjective , but guess what I am listening to now ?
OS
But about AX I am missing something.I remember that you could choose between TMC and CMC from the schematics (my one has the same, I just did not make a note on the sch.) but I did not remeber about TPC. Do you have and updated schematic, after 1.3 ( a link on the thread would do)? Did you put a resistor to set the static gain? I do not mind about PCB.
Thank you
Last edited:
since i am also an sziklai friend ( i will actually get him a beer if and when we meet ) here is my two cents ..
warphedale used a similar circuit for semipro use rated at about 150W @8R inside a amped speaker...
topology was simple LTP / CCS /dual VAS /1388-1859 drivers/clasic VI limmiters /1492 3856 outputs (4 totally)
i meet with this circuit in 2006 and to my surprize there was no emmiter nor collector balast resistors and found my shelf thinking that circuit like that is obviously not safe
amped speaker like the above is working absolutelly problem free from 2006 till today withouth the singlest sign of any trouble ... keep in mind that semi PRO 15" 2 way speaker in rental application is the speaker that is getting rented almost 3 times a week and means that obviously has been pushed around seriously alll those years ...
weird isnt it ?
warphedale used a similar circuit for semipro use rated at about 150W @8R inside a amped speaker...
topology was simple LTP / CCS /dual VAS /1388-1859 drivers/clasic VI limmiters /1492 3856 outputs (4 totally)
i meet with this circuit in 2006 and to my surprize there was no emmiter nor collector balast resistors and found my shelf thinking that circuit like that is obviously not safe
amped speaker like the above is working absolutelly problem free from 2006 till today withouth the singlest sign of any trouble ... keep in mind that semi PRO 15" 2 way speaker in rental application is the speaker that is getting rented almost 3 times a week and means that obviously has been pushed around seriously alll those years ...
weird isnt it ?
Dear Sakis, I kwow you by fame here, and I know that you like Sziklai (add another beer on my account), so I was somehow especting your post .
I hope you are speaking about the JAS 200 (double pair circuit).
I also believe that in the amplifier that you have seen, the OP transistors are carefully selected to have the same gain (at least on the same side), sotheir Re are not really needed . But what about the drivers' Re?
You see also that , since all collector are connected together, you can do WITHOUT insulating the BJTs, provided that you insulate the Heat sink itself from the chassis. This should bring some thermal transfer adavantage.
BTW could you be so kind to publish (or PM me) such Warfedale schematic?
Thaks a lot for your comments
I hope you are speaking about the JAS 200 (double pair circuit).
I also believe that in the amplifier that you have seen, the OP transistors are carefully selected to have the same gain (at least on the same side), sotheir Re are not really needed . But what about the drivers' Re?
You see also that , since all collector are connected together, you can do WITHOUT insulating the BJTs, provided that you insulate the Heat sink itself from the chassis. This should bring some thermal transfer adavantage.
BTW could you be so kind to publish (or PM me) such Warfedale schematic?
Thaks a lot for your comments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.