Thats a crazy price, but dont worry a design in the mold like here can sound even better. The main diffrence is the dartzeel has the 2 input transistors forming a diamond buffer that overcomes thermal problems and it doesnt use feedback around the outputstage (older version). The TSSA is different in that it doesnt have outputstage but the SSA can be as good as any dartzeel for a 100th of the price. Class A is obviously better but one can use SSA in AB and use dual feedback circuit giving very high openloop bandwith (greater than 20 khz) and lower distortion through output. I actually suspect that the dartzeel designer has maybe done this with this latest version but Id have to sim to get a better idea of what is different about the outputstage according to THD figures obtained. Two loops one from vas, one from output section to lower THD just a little, what you get is pretty close to tube sound without the overpowering sweetness and slow bass tube amps produce. Something tells me the dartzeels are heavily influenced around 4 commercial available designs from the late 70s to around late 90s that used very similar schemes but without the diamond input buffer transistors.
Last edited:
I think that this supports my idea that all links/items/surrounding’s in the reproduction should/must be seen as music instruments them self’s, and thus, if you introduce a beautiful instrument it may in fact better the sound, and what gets out may sound better than that what was put in. See also http://www.diyaudio.com/forums/analogue-source/154210-mpp-634.html#post3103164
Last edited:
I can agree to a certain degree with this, the thing is why do these effects with the amp and not the source which is first to reproduce, I think it would be easier to just have amps be wire with gain. The biggest reason tube amps made such a strong comeback lies with the gritty sound of CD, but if you use vinyl so called wire with gain amps sound just fine but not so much with CD.
When the recording is very good then the chain should be as transparent as possible, that is the ideal situation. Unfortunately a lot of music that i treasure is not recorded that well.
For a certain time i used a Cello Pallete in an attempt to get the best out of non ideal recordings. Ultimately i took it out of the system. It just consumed too much space and time. Now i prefer my system to be "musical". What that is is a bit hard to explain but it allows me to do a continued session of listening without adjusting all the time.
For a certain time i used a Cello Pallete in an attempt to get the best out of non ideal recordings. Ultimately i took it out of the system. It just consumed too much space and time. Now i prefer my system to be "musical". What that is is a bit hard to explain but it allows me to do a continued session of listening without adjusting all the time.
In Torino some years ago I witnessed a simple lesson. We were in the flat of a wealthy music lover who had a system by Cello which included the Palatte and which drove the big Audio Lab electrostatics. His taste was jazz. The effect, though not without room for criticism, was stunning. A pair of tube based mono preamps designed and built by a friend were inserted. Immediately the shortcomings of the earlier set up were staring us in the face. The sound we then heard was simply more credible, the dynamics were immediate and life like, they were solid and tangible.... in so far that sound can be tangible. There was a quality of pure magic when it was considered that this was sound reproduced by electronics.
Someone asked for the original pre to be put back in the system. The result was thin grainy unbalanced from top to bottom sound. We all breathed a sigh of relief when the tube mono pres were reintroduced. The difference in sound was equivalent to the average domestic higher end system being compared with a cheap transistor kitchen radio.
[Joachim knows this designer and also knows his design objectives.]
The point is that this experience gives me the basis on which to totally agree with FdW's assertion above regarding magical individual pieces of equipment totally transforming a 'system'. In this case the original host system was - in any terms - superb of its type. The insertion of a preamp change took it into the realms of unbelievably good!
Someone asked for the original pre to be put back in the system. The result was thin grainy unbalanced from top to bottom sound. We all breathed a sigh of relief when the tube mono pres were reintroduced. The difference in sound was equivalent to the average domestic higher end system being compared with a cheap transistor kitchen radio.
[Joachim knows this designer and also knows his design objectives.]
The point is that this experience gives me the basis on which to totally agree with FdW's assertion above regarding magical individual pieces of equipment totally transforming a 'system'. In this case the original host system was - in any terms - superb of its type. The insertion of a preamp change took it into the realms of unbelievably good!
Dual mono, spacey and logical modules layout inside the case. Heatsinks looking capable of class A heat dissipation, only the transformers seems a little too small but that's probably just visual, VA rating is sufficient I suppose. All mentioned is just basics, major feature is an amp's topology and the sound quality achieved, as told, congratulations again to an expert as you are Sonny.
Lately in the thread diamond buffer (DB) as an input and current feedback was mentioned, so just few words about it. Before SSA I also tried to use DB as an input stage and after some critical measurements - instability, I realize that it has the same problem as LTP - two active devices too more in the signal path, this time at the input signal, regarding feedback signal at LTP's. The best stability, speed, bandwidth and therefore sound quality were achieved when modulating the same active device with input and feedback signal, that's why I still stick to this same input stage topology at SSA, TSSA and CSA, only that later two have less problems with tempco/bias conditions, so easier to handle. Sonny your previous amp has DB input stage and similar next stage, so the logical question would be, how do they compare regarding SQ?
BTW as seen on the pic, today I soldered all WIMA decoupling capacitors on both CSA PCB channels, 26 is their count number on single PCB as they are present on all critical points in the layout.
Attachments
Stability with two stages only is really remarkable, I remember in the testing phase when I used two stages only, no compensation cap was needed at all. BJT as front-end follower has more gain (also more OLG) so it is more suitable as a VAS transistor or CFP input, than mosfet. Extra gain dB (CLG subtracted from OLG) obviously and expectedly reduce THD.
Now currently by CSA, even with 28 transistors present on PCB, it is still three stages amplifier only, maintaining superb stability, achieved with minimal compensation needed.
It needs compensation. That is a fact.
On the new board, i have added 47pF miller cap across KSA1381 + KSC3503.
At certain source impedances it was oscillating when i was changing it from Renesas 2SK216 + 2SJ76. The 47pF matches roughly the Cgd value.
It is a concern with CFP BJT pair as they have there own pole. It gives more gain and reduces re (emitter equalint resistance).
An easy way to get more gain (lower THD) and reducing the output impedance is to use ECW20N20-Z/ECW20P20-Z or ALF16N16W/ALF16P16W.
This way the class A current can be set higher as well as they have a lower K/W.
On the new board, i have added 47pF miller cap across KSA1381 + KSC3503.
At certain source impedances it was oscillating when i was changing it from Renesas 2SK216 + 2SJ76. The 47pF matches roughly the Cgd value.
It is a concern with CFP BJT pair as they have there own pole. It gives more gain and reduces re (emitter equalint resistance).
An easy way to get more gain (lower THD) and reducing the output impedance is to use ECW20N20-Z/ECW20P20-Z or ALF16N16W/ALF16P16W.
This way the class A current can be set higher as well as they have a lower K/W.
It was demonstrated on the fetzilla thread that actually more OLG can be achieved with a mosfet as VAS than with a BJT.
The i/p stage and VAS are in anti phase and any current leakage between them ( that BJT's draw ) results in in a loss of OLG which dominates over any extra gain caused by a BJT have higher transconductance. Because of this the fetzilla with it's Jfet i/p needed a mosfet VAS to give enough gain for the design to sound good.
However in Sonny's design here it is not so straightforward because there we have a BJT / MF darlington pair so I'm not sure how this would impact the explanation above. But with any 3 stage design I believe a MF VAS will give higher OLG.
The i/p stage and VAS are in anti phase and any current leakage between them ( that BJT's draw ) results in in a loss of OLG which dominates over any extra gain caused by a BJT have higher transconductance. Because of this the fetzilla with it's Jfet i/p needed a mosfet VAS to give enough gain for the design to sound good.
However in Sonny's design here it is not so straightforward because there we have a BJT / MF darlington pair so I'm not sure how this would impact the explanation above. But with any 3 stage design I believe a MF VAS will give higher OLG.
Last edited:
well, with a simple version of TSSA I am playing around with in spice changing the O/P VAS stage driver transistor from BC560 etc to ZNV3306 reduces OLG from 44.45 to 44.07 !
So no real effect in this case and because the mosfet had more voltage drop BJT seems like the best choice in this circuit.
So no real effect in this case and because the mosfet had more voltage drop BJT seems like the best choice in this circuit.
This is near always the case, even with any LTP circuit, mosfet will have higher gain, but that is not such a good thing as you might think. As the mosfet has much higher capacitances it will show much higher high frequency distortion (usually from 2 Khz and up). The only benefit of mosfet vas is only a very small lowering of low frequency distortion. A plain lone mosfet will never outperform a BJT as Vas, even when cascoded and only if you know some groovy tricks its can possibly outgun a BJT.
Last edited: