If you have lateral MF o/p devices they can be driven with 10mA in VAS and in this case a zvn3310 can be used with quite low i/p capacitance.
In Fetzilla with Jfet i/p device the cct measured & sounded better with MF VAS.
So your statement above is not true in all topologies
If you have lateral MF o/p devices they can be driven with 10mA in VAS and in this case a zvn3310 can be used with quite low i/p capacitance.
In Fetzilla with Jfet i/p device the cct measured & sounded better with MF VAS.
So your statement above is not true in all topologies
I doubt it, very much, for one no real measurements were shown, only sims but even as such this should become obvious if you simmed and compared the THD at 20khz and not just at 1 Khz, the mosfet will lose above 1 Khz, try it. What it sounds like is a different matter. In all topologies is true because this behaviour is same for any common emitter circuit using mosfet be it a voltage feedback amp or CFB amp.
I checked Fourier with 10Khz fundamental with BJT & MF VAS in fetzilla in spice
MF 2nd HD = -76db 3rd HD = -87db OLG = 64db
BJT 2nd HD = - 62db 3rd HD = -77db OLG = 57db
So I would still recommend MF as VAS and have often wondered why not many people do not choose this option - I wonder have they tried and not liked or is it dogma ???
cheers
mike
p.s. The ZVN3310a has particularly low i/p capacitance and sounded best of all devices tried my myself & greg peters
MF 2nd HD = -76db 3rd HD = -87db OLG = 64db
BJT 2nd HD = - 62db 3rd HD = -77db OLG = 57db
So I would still recommend MF as VAS and have often wondered why not many people do not choose this option - I wonder have they tried and not liked or is it dogma ???
cheers
mike
p.s. The ZVN3310a has particularly low i/p capacitance and sounded best of all devices tried my myself & greg peters
Last edited:
I'm currently designing an amp with Mosfet VAS (2SK216/2SJ79).
Picture can be seen in post#8 from thread:
http://www.diyaudio.com/forums/solid-state/217528-low-power-mosfets-complementary-pairs.html
A bjt as VAS outperforms the FET, but only a tiny bit in THD value. 2th and 3th harmonic are the same in level. The 5th is -220dB with the BJT and -205dB with the Fet as VAS. 6Th harmonic is not existing with 1Vrms output (normal listening level).
The FET as VAS is winning in Open loop bandwith(10kHz) and number of higher harmonics. ==> I'm definetely going for the FET's this time.
My 2cents.
Last edited:
Sonny, opinion please.
Regarding difference in diamond buffer contra the (t)ssa frontend....
The diamond buffer is easier to stabilize, All for transistors needs to be in close proximity. If this is the case and the emitter resistors are identical, i works like a charm.
The (t)ssa does not need those resistors. Those resistors will influence the step response.
Of course there is one layer transistors more in the diamond buffer but they only act as a buffer to the transistors doing the actual error correction.
The TSSA is stable but it also has ac coupling caps in series with the feedback resistors to ground.
,
The diamond buffer does not need them, nor does the SSA stage. But the SSA needs a lot of house keeping to keep the current and voltage drift in place.
Onto the difference between the TSSA V1.6, CFB 1xECX10 V1.1, Mirand A1 V1.1..
The CFB 1xECX10 V1.1 is DC coupled with SERVO. The frontend current generators are not as fine as in the TSSA V1.6..
This amp is fine... We are talking copland Level, but there is compared to the TSSA V1.6 a little haze over the picture it paints. The sound stage is not as wide as with the TSSA V1.6
The Mirand A1 V1.1 is and old design with better current generators and hawksford cascodes on the current mirrors. This amp is very stable and could run without servo.
In the comercial section under SITO Audio you can find distortion specs.
I actually believe it is on the level of the TSSA but it is also more expensive to build (3x) . It could need an upgrade to the current generators in the TSSA....
It has more grip and control in the base, midrange..And in hififorum.nu it was described as being playing endlessly deep....
I have not rejected the diamond buffer or the TSSA frontend...
One thing is for sure ...i have a bigger brother to the TSSA and MIRAND in the fire right now and it will be a mix.
Still need to make the final decision of the VAS stage.
It will not be a cheap solution at all.
Benson Im afraid those models you have for those mosfets are for all practical purposes useless. There will be a fair amount of distortion over a BJT at high frequencies. The mosfets maybe win on open loop bandwith only because the miller effect starts to be felt and they rapidly lose their gain at these frequencies and at this stage have much lesser gain. Less gain - higher bandwith but less feedback - higher distortion. This effect usually starts at 2-3 khz.
Sonnya,
I have 2 PCs of 2x 40V AC - 9,3 A sec HR toroids + - 55V DC and would like to give a try Your TSSA V1.6 amp.
What is the max DC voltage accepting by the amp and would a 5 pairs of Exicons be sufficiently driven by the current VAS as is? or need to be made some minor changes within the proposed latest all fet VAS - OPS schem. You post here ?
Thank You
Thx Sonny,
Not sure if you listened to your TSSA with higher bias and what you thought of the sound if you did, but I would suggest that with an amp with such uncommonly low OLG & therefore low FB, there will be an unusually high improvement when it is run in class A.
I would certainly want to try this type of design with at least 1A bias and I would use + / - 25V rails to allow this - distortion figures will be significantly lower and I imagine the sound would be even better than you describe.
If you have tried this and didn't like the sound I would be somewhat perplexed
cheers
mike
Not sure if you listened to your TSSA with higher bias and what you thought of the sound if you did, but I would suggest that with an amp with such uncommonly low OLG & therefore low FB, there will be an unusually high improvement when it is run in class A.
I would certainly want to try this type of design with at least 1A bias and I would use + / - 25V rails to allow this - distortion figures will be significantly lower and I imagine the sound would be even better than you describe.
If you have tried this and didn't like the sound I would be somewhat perplexed
cheers
mike
You are right Mike. It does sound "better" at higher bias. But i settled for 0.6A for reliability reasons.
The temperature rise was just to high for my taste.
Secondly 0.6A and 32V with an heatsink of 0,4K/W + isolation (0,4K/W) + 1K/W for the the mosfets will give you a rise of 42degrees over room temperature.
On a hot day in south of europe or near equator will give you an margin of 45 degrees.
If we go for 1A at 32V it would leave a margin of 15 degrees. That will wear out the components fast.
I wanted to make something that runs at your guys systems 15 years or more out in the future.....
Secondly the mosfet has a voltage drop of about 5V. with this voltage it will leave you with 25W. A bit low for my taste. But it is not a problem.. just use a 2x18VAC transformer... It will run happely with it.
With 32V you will have ~40W @ 8Ohm
I just want to show my appriciation on the guys here in this forum.
Until now i have got orders 7 sets. Is is not a cheap kit, and you will not be dissapointed ... All the ones that have ordered them.
As Mikelm have pointed out. If you can live with less power (25W@8R) you can raise the idle to full class A.
Until now i have got orders 7 sets. Is is not a cheap kit, and you will not be dissapointed ... All the ones that have ordered them.
As Mikelm have pointed out. If you can live with less power (25W@8R) you can raise the idle to full class A.
Sonny I wish you very gooood sales on TSSA, you've deserved it.
It would be nice to hear SQ feedback from the buyers. You can direct them to this thread so we'll all get an impression, images etc.
Regards Andrej
...... my contribution
regards Alex.
WOW perfect take out of Sonny's sch.
All time classic, old school.
Now DIY-ers can test & taste TSSA with through hole parts too.
Thanks Alex
WOW perfect take out of Sonny's sch.
All time classic, old school.
Now DIY-ers can test & taste TSSA with through hole parts too.
Thanks Alex
I could be from, since i have some 2SJ162/2SK1058 ond hand. BC550/BC560 could be use insted of bc850/560 seems like to be the smd brother of previous on. I have 300va 2x25vac trafo and some 22.000%f/40v screw caps. This board fit perfect for diy use : single layer => home etching....
Marc