It could be even better......
The reality is that unless you had both color bands to listen to you won't know the difference and be happy with the way the amp sounds.
The reality is that unless you had both color bands to listen to you won't know the difference and be happy with the way the amp sounds.
Fab or anybody else,
2 questions.
1) What is the reasoning for matching BJT’s Q1 and Q2 more closely in the input stage than Q3, Q4?
2) Is it possible to link out the input capacitor (C2) if there is no DC from our preamp or will the DC offset in the output be too high?
Thanks,
Anand.
2 questions.
1) What is the reasoning for matching BJT’s Q1 and Q2 more closely in the input stage than Q3, Q4?
2) Is it possible to link out the input capacitor (C2) if there is no DC from our preamp or will the DC offset in the output be too high?
Thanks,
Anand.
Last edited:
Hi
1) CFP gain Beta is Bq1*Bq3+Bq1 thus q1 is a little more contributing. However, the matching could be also closer than specified for Q3/Q4....I think in practice that most of you will select a closer match than specified anyway...😉 matching percentage value is not absolute and for reference only otherwise everybody would ask... It is not that important either.
2) yes C2 can be bypassed with good Input stage matching. You can mesure DC at input with input floating. Only if you have very low DC that you can connect a pot directly at entry of amp without a cap. Since USSA-5 has feedback caps then there is no amplification of DC present at input. Thus DC out = DC in + DC offset. If DC in is very small at input then there is no issue in normal operation. The drawback of removing C2 is if you have DC thump from preceding equipment such as a preamp... also it no more follows the "rule" of lower cutoff frequency of feedback than cutoff frequency of input... but the latter "rule " is of no real concern unless you have a phono stage which can create rumble (very low frequency).
Fab
1) CFP gain Beta is Bq1*Bq3+Bq1 thus q1 is a little more contributing. However, the matching could be also closer than specified for Q3/Q4....I think in practice that most of you will select a closer match than specified anyway...😉 matching percentage value is not absolute and for reference only otherwise everybody would ask... It is not that important either.
2) yes C2 can be bypassed with good Input stage matching. You can mesure DC at input with input floating. Only if you have very low DC that you can connect a pot directly at entry of amp without a cap. Since USSA-5 has feedback caps then there is no amplification of DC present at input. Thus DC out = DC in + DC offset. If DC in is very small at input then there is no issue in normal operation. The drawback of removing C2 is if you have DC thump from preceding equipment such as a preamp... also it no more follows the "rule" of lower cutoff frequency of feedback than cutoff frequency of input... but the latter "rule " is of no real concern unless you have a phono stage which can create rumble (very low frequency).
Fab
Last edited:
How many might one need since they are unmatched sets?
Presumably you posted the GB meaning someone has them for sale there?
Presumably you posted the GB meaning someone has them for sale there?
I mounted a USSA-5 without having paired 2SK2013 / 2SJ313 and it poses no problem, if you match these transistor is better but not necessary and it is the same for power mosfets.
Fab can confirm this answer I think.
Have you read the post that link points?
anybody know if this one is reliable source?
TOSHIBA MOSFET 2SK2013 2SJ313 Pair 2-10R1B_Transistor_Analog Metric - DIY Audio Kit Developer
TOSHIBA MOSFET 2SK2013 2SJ313 Pair 2-10R1B_Transistor_Analog Metric - DIY Audio Kit Developer
I do not know if at Analog Metric they are authentic but I bought mine at Audiowind-2005 and they were genuine.
2SJ313-Y and 2SK2013-Y TOSHIBA Audio Power MOSFET | eBay
2SJ313-Y and 2SK2013-Y TOSHIBA Audio Power MOSFET | eBay
Hi
See post 75.
Also Nicmac members.
Regarding matching it is only as an insurance that they will have about the same VGSoff at 20ma. The datasheet indicates a variation from 1.8 to 2.8 volts which is too wide. VGSOff matching at 20ma of +/-10% should be adequate. If you want to have the exact same performance between left and right channel you may want a quad matched to the same tolerance. Not doing so for tolerance of the same channel (left or right) will work anyway but may create more THD and I have not tested the sound result to comment.
Fab
Last edited:
On the build instructions it states:
We need to increase current to about 20ma for IM3, thus:
o RM3 = VGSP / 20ma = ________
My question is, where did the 23.2mA value come from that ggetzoff has highlighted on his build sheet? Is this a measurement or calculation that is not presented?
Thanks,
Steve
Hi
20ma driver current at 1A bias during test phase.when you increase to about 1.3A as expected for final bias then Mosfet current increases to about 23ma or so. Exact value for about 1A bias not important but choose value which allows standard resistors to have same current in positive than in negative portion of circuit. For Ggetzoff specifically he told us we would use 1A as final bias (he has high efficiency speaker so he will not use 25Wrms power) so that is why we targeted to 23ma for Mosfet driver current during its resistors determination phase. This is a special situation. Having said that avoid lower current than about 20-22ma for driver current when final output bias is set.
Fab
20ma driver current at 1A bias during test phase.when you increase to about 1.3A as expected for final bias then Mosfet current increases to about 23ma or so. Exact value for about 1A bias not important but choose value which allows standard resistors to have same current in positive than in negative portion of circuit. For Ggetzoff specifically he told us we would use 1A as final bias (he has high efficiency speaker so he will not use 25Wrms power) so that is why we targeted to 23ma for Mosfet driver current during its resistors determination phase. This is a special situation. Having said that avoid lower current than about 20-22ma for driver current when final output bias is set.
Fab
Last edited:
I'm busy collecting parts for one of these. Can you please write some details about the rating of your trafos? I've got some nice 500VA Selectronic R-cores but at 24v I think they might be just a bit too big?
Fab insists in the manual on a maximum voltage of +/- 32 Vdc, so I think your transformers will not fit for this amplifier.
Thanks
So we will need no more than 22v trafo so that peak after rectification is no more than +/- 32v
Right?
So we will need no more than 22v trafo so that peak after rectification is no more than +/- 32v
Right?
It should be confirmed by more technical people than me in this area but I will say that a transformer of 22 Vac gives after rectification and filtering a voltage of 30 Vdc, so it should be a priori.
Had these things packed away in a box so long that I had forgotten exactly what they were. Got them for a Gainclone project that I have just never got around to.
Anyway dug them out of the back of the cupboard because I wasn't sure about their rating when I last wrote ...
BINGO!!
they are 21v trafos - so they should fit comfortably below the ceiling under load after rectification/filtering
Gonna need a soft start! with 500va per channel
Yes that should be fine and 500VA should give you a very efficient PSU while you use a C-R-C configuration. However with about +/-27vdc , you may have to limit your bias to about 1A (about 54W dissipation per channel) -instead of 1.2A or more - so your heatsink (depending on size) temperature does not get too high...
Limiting bias to 1A will reduce class A power (for the same load) while higher voltage will increase class AB power. But 1A should still get you close to 20W or more class A power (8 ohms) because of the square law effect of the output Mosfet. Also while reducing to less than 1.2A bias you will need to target more about 23-25ma or so during ajustement instead of 20ma for the driver stage (as was done for Ggetzoff amp).
By the way the sound of USSA is quite different than a Gainclone.
Fab
Limiting bias to 1A will reduce class A power (for the same load) while higher voltage will increase class AB power. But 1A should still get you close to 20W or more class A power (8 ohms) because of the square law effect of the output Mosfet. Also while reducing to less than 1.2A bias you will need to target more about 23-25ma or so during ajustement instead of 20ma for the driver stage (as was done for Ggetzoff amp).
By the way the sound of USSA is quite different than a Gainclone.
Fab
Last edited: