What do you want to use them for? Also, will they have a sub woofer to augment bass, or standalone? How much bass extension do you want? Are you looking for small pair of speakers to use around the house with an mp3 player? All these things will help determine an ideal cabinet. I have made several different designs ranging from a mini Karlsonator, a DCR, an XKi, a sealed, a MLTL, Cornu, Nautaloss, etc... To get some ideas of what you can do with a 3.5in full range driver look in the index (post 1) of this thread (mostly TC9FD but RS100 can be used in many cases with slight mods):
http://www.diyaudio.com/forums/full-range/223313-foam-core-board-speaker-enclosures.html
If you also have an 8in driver, I would suggest a FAST similar to this:
http://www.diyaudio.com/forums/full-range/273524-10f-8424-rs225-8-fast-ref-monitor.html
For a small standalone speaker to play music from a phone with a small class D amp, the 0,4x Karlsonator is excellent:
http://www.diyaudio.com/forums/full...sonator-0-53x-dual-tc9fds-16.html#post3722988
For a great set of PC monitors, a DCR can work very well:
http://www.diyaudio.com/forums/full-range/252627-viva-la-vifa-curvy-cabinet-dcr-tc9fd.html
Those are just some ideas.
Typical BSC for a small 3 in to 4in driver in 4 pi space is 1mH choke in parallel with 5R to 10R, adjust to taste. Youi can use cheap little 1amp 1mH ferrite chokes normally used for class D amp or SMPS filtering. Works fine and save $10 from having to get an air core choke.
Last edited:
F5 Juma edition (2SK2013/2SJ313 outputs)
The long awaited sequel to this amp comparo is now the legendary Pass F5, now using Juma's mod which utilizes paralled small Toshiba 2SK2013/2SJ313's which have superb linearity as they were designed for audio rather than the IRFP240/9240's which were designed for power supply switching.
The F5 Juma is described here, but briefly the schematic looks like this:
It's very simple with a pair of little JFETs driving a whole gang of MOSFETs directly. I am using 4 pairs vs the 3 pairs shown in schematic. It has been confirmed by other builds that even 5 pairs can be driven and sound good.
My build is in 3d P2P, no veroboard, no PCB's. I am using 0.22R source resistors as I did not have any 0.33R non-inductive 2W units handy. I added 2200uF rail smoothing caps and a 10R-100nF Zobel on the output. Also, I am running it with 20dB gain vs 14dB design gain to avoid having to need a pre-amp (I am using only a single 100R feedback per leg vs 2 in parallel). My amp is running 1.45A overall bias with individual MOSFETs running between 300mA and 370mA ea (no, I could not afford to match them). Power supply is 25v rails fed from 400VA with CRC filter (same as used on M2). Here is the amp built up:
Here it is in action with cooling fans:
Here is the measured frequency response in comparison to the M2 and F9HV:
Here is the measured HD - note how uniform the components are and how low the bass distortion is - it drives the woofer with great control and has less HD below 100Hz than any other amp tested in this thread (IIRC):
Here is the measured phase:
Here is the measured impulse and step response:
Finally sound clips are below. As usual, change extension from .asc to .mp3 to listen.
Let me know what your listening impressions of this amp are. I am debating whether or not to build the second channel.
The long awaited sequel to this amp comparo is now the legendary Pass F5, now using Juma's mod which utilizes paralled small Toshiba 2SK2013/2SJ313's which have superb linearity as they were designed for audio rather than the IRFP240/9240's which were designed for power supply switching.
The F5 Juma is described here, but briefly the schematic looks like this:
It's very simple with a pair of little JFETs driving a whole gang of MOSFETs directly. I am using 4 pairs vs the 3 pairs shown in schematic. It has been confirmed by other builds that even 5 pairs can be driven and sound good.
My build is in 3d P2P, no veroboard, no PCB's. I am using 0.22R source resistors as I did not have any 0.33R non-inductive 2W units handy. I added 2200uF rail smoothing caps and a 10R-100nF Zobel on the output. Also, I am running it with 20dB gain vs 14dB design gain to avoid having to need a pre-amp (I am using only a single 100R feedback per leg vs 2 in parallel). My amp is running 1.45A overall bias with individual MOSFETs running between 300mA and 370mA ea (no, I could not afford to match them). Power supply is 25v rails fed from 400VA with CRC filter (same as used on M2). Here is the amp built up:
Here it is in action with cooling fans:
Here is the measured frequency response in comparison to the M2 and F9HV:
Here is the measured HD - note how uniform the components are and how low the bass distortion is - it drives the woofer with great control and has less HD below 100Hz than any other amp tested in this thread (IIRC):
Here is the measured phase:
Here is the measured impulse and step response:
Finally sound clips are below. As usual, change extension from .asc to .mp3 to listen.
Let me know what your listening impressions of this amp are. I am debating whether or not to build the second channel.
Attachments
Last edited:
Hi xrk971,
many thanks for this interesting measurement,
Not sure, but I think there is a slight gap between this updated measurement of FH9 and the measurements on post #1, visible between 1kh and 1.8khz. maybe it is only a printing issue.... Have you change something on the PSU?
Except at 1.5khz, on your system, all of these amplifiers look quite similar in term of performance.
At 1.5khz FH9 looks better.
If my understanding is correct, Impulse and step response are not better than other amplifiers you measure... Quasi mofset looks better on this point.
many thanks for this interesting measurement,
Not sure, but I think there is a slight gap between this updated measurement of FH9 and the measurements on post #1, visible between 1kh and 1.8khz. maybe it is only a printing issue.... Have you change something on the PSU?
Except at 1.5khz, on your system, all of these amplifiers look quite similar in term of performance.
At 1.5khz FH9 looks better.
If my understanding is correct, Impulse and step response are not better than other amplifiers you measure... Quasi mofset looks better on this point.
Hi xrk971
Similarly to the F5-Juma, you can also try an F5 type amp with other more powerful audio mosfets designed also for audio such as 2SK1530y/2sj201y as a single output pair using an already available F5 pcb on the net. I believe that EUVL member has also experimented with these mosfet in an F5x topology if I am not mistaken. The transconductance of these mosfet is as high as an IRFP240/9240 but since the threshold voltage is lower than the resulting damping factor is reduced and provide a favourable sound....Also, the price for a single pair not requiring matching is lower than multiple matched 2Sk2013//2sj313 ....but the sound might be different also...
See at this link:
F5 avec la recette .....USSA
Fab
Similarly to the F5-Juma, you can also try an F5 type amp with other more powerful audio mosfets designed also for audio such as 2SK1530y/2sj201y as a single output pair using an already available F5 pcb on the net. I believe that EUVL member has also experimented with these mosfet in an F5x topology if I am not mistaken. The transconductance of these mosfet is as high as an IRFP240/9240 but since the threshold voltage is lower than the resulting damping factor is reduced and provide a favourable sound....Also, the price for a single pair not requiring matching is lower than multiple matched 2Sk2013//2sj313 ....but the sound might be different also...
See at this link:
F5 avec la recette .....USSA
Fab
Thanks for the tip Fab. Where do you get those MOSFETs from? What's your impression of how the F5 Juma mod sounds? Truth be told, I did not match at all and took what I got. I did swap on out two transistors that ran extra hot (3x current of the others), and I know many will be incredulous that these MOSFETs are even useable as they were only $4.35 for 5 pairs. But seems to work and I like the sound. The measurements aren't half bad either. I have burned them in over a day now and still working with stable bias current near 1.5amp for the whole set.
Last edited:
If you paid 4.35$ for 5 pairs of 2SK2013/J313, they're garanteed to be fake and this might not give you the sound of genuine ones. I'm just saying but it is too good to be true, especially for parts like these ones that many wants and hard to find. Chances are that a genuine pair would cost around 10$+ and for a single pair only.
Anyways, I do hope they're genuine and that your enjoying the sound of them. There's a test you can run to make almost 100% sure you have genuine stock, please see here:
Amplificateur USSA-5 : évolution de la conception de la version 4 - Page 4
You can google translate the page but I believe one person has an excel calculator sheet and another a program. You can take measurements and you will see if it matches the specs properly. The test was made exactly for these Mosfets.
Ciao!
Do
I see tables there for currents and voltages. What's the test circuit look like? I agree and doubt genuine. I wonder what MOSFET is actually hiding inside then? The casing and lettering are quite well done in any case. I would like to test them just to see how far off it is from real ones. If real ones indeed were $10/pair (almost as much as a set of real sanken ring emitter BJTs) that would be a deal breaker and I simply look elsewhere for another amp design. Life is too short to fuss with unobtaanium boutique transistors.
As someone on this forum told me: just because they sound good and measure well doesn't t mean they are real. But then, if they sound good and measure well, what am I complaining about?
As someone on this forum told me: just because they sound good and measure well doesn't t mean they are real. But then, if they sound good and measure well, what am I complaining about?
Last edited:
You may be right....
as I am a real noob in class AB DIY amplifier, I didn't want to invest a lot, and I bought all my components at alliexpress / china ebay.
I can say You are right:
resistor marked 1% are not 1% but 5%
(and ?) measurements for transistors are... not well aligned with specifications....
I can also say:
I never have the luck to listen a 10.000$ amplifier....
But for the money I paid (few euros, maybe 2 beers
), the state of the system I built (definitely not well build prototype) the music I am currently listening is far far better than all the middle range amplifier I have/had and my friends have.using chinese components made me happy and would like to improve my system, build better amplifiers... etc...
http://www.diyaudio.com/forums/soli...imate-fidelity-amplifier-872.html#post4846682
My 1% resistors have been correct. They are laser trimmed so generally accurate. The electrolytic caps have always been measured at the bottom of their rating. So a 10% 2200uF will be no more than 2000uF. I have had good luck with transistors except one set which were fake and I got a refund.
But I have to agree that Chinese components have allowed me to make many amps that sound very good for very little money.
But I have to agree that Chinese components have allowed me to make many amps that sound very good for very little money.
You are welcome. These mosfet can be bought with confidence from here with no matching required thus not much that expensive for the quaranteed quality you get...
http://www.diyaudio.com/forums/vend...uding-some-very-rare-toshiba-transistors.html
I have another reliable source from Ampslab as well.
I have only compared between F5 original layout and mine but did not have the chance to compare with Juma...
For the measurement you only have to do a voltage divider at gate of mosfet using a pot and put a a source resistor and draw the ID-vgs curve against the data sheet. Once done you draw the transconductance YFS-ID and compare with data sheet. The data sheet also provides good tips for the genuine package (date code) and the picture about the thickness at different areas and small holes, etc...
Good luck with either mine or Juma or both !
Fab
Last edited:
They are better tasting not fried...
Look at the single pair 2sk1530/sj201 solution and it is not very expensive compared to multiple pairs 2sk2013/sk313 in the F5...
On my side I use the 2sk2013/sj313 more for driver than output...
Fab
Last edited:
Second set of Comp Quasi amps (Dacz layout):
Measurements compared to first Quasi here:
http://www.diyaudio.com/forums/soli...mentary-mosfet-amplifier-151.html#post4859870
Measurements compared to first Quasi here:
http://www.diyaudio.com/forums/soli...mentary-mosfet-amplifier-151.html#post4859870
hi Xrk
As these Class D .... http://www.diyaudio.com/forums/class-d/275915-200w-irs2092-amp-20-a-36.html#post4640601 ...... compared to class AB currently mounted?
there is not much difference?
Thiago
As these Class D .... http://www.diyaudio.com/forums/class-d/275915-200w-irs2092-amp-20-a-36.html#post4640601 ...... compared to class AB currently mounted?
there is not much difference?
Thiago
Oh yes. I was one of the first on this forum to build a TPA3118 amp from scratch. I used TPA3118/16 for years as my main active system amps. They work very well for what they are but are limited in their dynamics and bass authority. They lack the transient power to keep up with a good class AB amp in this forum like the Quasi by Ranchu/Aksa or the FH9 or CFH7. Those amps also have lower harmonic distortion at full power as well. Once you build one and hear one the differences are very clear. I pretty much have stopped using class D amps.
This was before cheap boards were ever available from China.
http://www.diyaudio.com/forums/clas...ding-them-everything-comes-6.html#post3413886
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.