F3 Boards
Thanks for your suggestion Peter. Fortunately, member 'bluemartini' has offered to part with a spare set he purchased before he built his F3. If both of your quick responses to my call for info. are any indication of the level of interest and support that exists in this Forum, building the F3 should be more interesting and enjoyable than going it alone. I'll share my progress once the F3 project is underway.
Thanks for your suggestion Peter. Fortunately, member 'bluemartini' has offered to part with a spare set he purchased before he built his F3. If both of your quick responses to my call for info. are any indication of the level of interest and support that exists in this Forum, building the F3 should be more interesting and enjoyable than going it alone. I'll share my progress once the F3 project is underway.
Re: Adjusting the "ghost load"
Hmm, finally got a F3 installed on heatsink.
I have been primary looking at the Z9 article, R6-8 is .7v and set P1 for Q2 drain of 23v. Reading the above, not sure what to do with P2, I have no scope...... Adjust by ear?
I may have missed a good post on it somewhere.
Oh ya, and I think I have to move the FETs around on the sink, Q2 casing is running at 90c
I think my safe point is 65c. I am running mica and grease too.
Snokker said:Fellow F3 builders,
In the original setup of my F3 I calibrated P2 to nearly a minimum value, setting the total distortion on the output at the lowest possible level. (load= 8 ohm + 1mH in series + 1uF parallel; 1kHz@ 1W)
Now I've been listening to different speakers I'm not sure if this is the way to go. Reducing the ac feedback of the output on the setting of the current source seems to sound more natural and less thin (but maybe less controlled...)
Can anyone share his/her
BTW the work on a "F6" a two cascoded LU1014D per channel version of the F3 is suffering from heatproblems (even in wintertime....)
Hmm, finally got a F3 installed on heatsink.
I have been primary looking at the Z9 article, R6-8 is .7v and set P1 for Q2 drain of 23v. Reading the above, not sure what to do with P2, I have no scope...... Adjust by ear?
I may have missed a good post on it somewhere.
Oh ya, and I think I have to move the FETs around on the sink, Q2 casing is running at 90c
I think my safe point is 65c. I am running mica and grease too.Suggest you go by the Cct First Watt, F3 R0 (11/04/05).
If the Q2 = Lu Jfet's cascode, and your rail after the Cmultiolier Q5 about 41 volts, then the drain on your Q2 should be close to 21 volts and it's source about 3.5 volts, so Q2 is sinking about 17 volts @ 1.7A = 29W,[As you measure 0.7 volts across R6,7,8 this shows you have about 1,7Amps thru the whole "string"] (which is about right)], and so that 90*C seems to indicate a problem somewhere.
Possibles that I've done - raised lip on tapped hole for hold-down screw, plastic insulator instead of mica (yeah, unbelievable!), the usual loose/ stripped screws, dirt under trannie, etc, etc
One thing you might checkm is the voltage at the gate of Q2, set by the P1 - should be about 8volts - perhaps this may be a problem.
Incidently, the quality of the C2 capacitors has a significant influence to the sound, along with C3 and C4.
In this amp, I've found that the F&T caps from "The Audio Capacitor Co" (in the UK) sound the same as the Mundorfs and the Jensens as power supply and o/p caps.
If the Q2 = Lu Jfet's cascode, and your rail after the Cmultiolier Q5 about 41 volts, then the drain on your Q2 should be close to 21 volts and it's source about 3.5 volts, so Q2 is sinking about 17 volts @ 1.7A = 29W,[As you measure 0.7 volts across R6,7,8 this shows you have about 1,7Amps thru the whole "string"] (which is about right)], and so that 90*C seems to indicate a problem somewhere.
Possibles that I've done - raised lip on tapped hole for hold-down screw, plastic insulator instead of mica (yeah, unbelievable!), the usual loose/ stripped screws, dirt under trannie, etc, etc
One thing you might checkm is the voltage at the gate of Q2, set by the P1 - should be about 8volts - perhaps this may be a problem.
Incidently, the quality of the C2 capacitors has a significant influence to the sound, along with C3 and C4.
In this amp, I've found that the F&T caps from "The Audio Capacitor Co" (in the UK) sound the same as the Mundorfs and the Jensens as power supply and o/p caps.
Re: Re: Adjusting the "ghost load"
Hi Teabag,
After a few months of fine-tuning my F3, distortion is really low (0,004%@1W-8Ohms) The funny thing is that the effect of changing the value of P2 has the most effect on the even (2nd) harmonics. The 3rd harmonic seems to have a basic level. Not the usual thing with NP current source amps...
Adjusting an F3 to maximum performance is simultaniously tuning R5, P1 and P2. The V values given in the F3 R0 schematic are a good start + P2 close to the the minimum value.
Nobody near you with a distortion analyser??? that would really help and gives insights of the characteristics of the LU1014.
(You could try to set up an input vs output differential like David Hafler did with his XL280 amp; mind the inverted output!!!)
In your case I would first solve the problem with Q2 (check voltages!). Did you swap the LU104's already? they really have huge tolerances and directly effect the setting of Q2.
Have fun
Tea-Bag said:
not sure what to do with P2, I have no scope...... Adjust by ear?
I may have missed a good post on it somewhere.
Hi Teabag,
After a few months of fine-tuning my F3, distortion is really low (0,004%@1W-8Ohms) The funny thing is that the effect of changing the value of P2 has the most effect on the even (2nd) harmonics. The 3rd harmonic seems to have a basic level. Not the usual thing with NP current source amps...
Adjusting an F3 to maximum performance is simultaniously tuning R5, P1 and P2. The V values given in the F3 R0 schematic are a good start + P2 close to the the minimum value.
Nobody near you with a distortion analyser??? that would really help and gives insights of the characteristics of the LU1014.
(You could try to set up an input vs output differential like David Hafler did with his XL280 amp; mind the inverted output!!!)
In your case I would first solve the problem with Q2 (check voltages!). Did you swap the LU104's already? they really have huge tolerances and directly effect the setting of Q2.
Have fun
jameshillj said:
Getting closer,
filed down tap hole a bit, sanded it, and then decided to put the berquest sil pads on, Things look better for sure. With 24v bias on Q2 drain I get 60c on Q2 get case, and Q3 gets to 65c. It is currently on the upper half of the heatsink, I am going to flip the mouting 90 degrees to get the heavy lifting fet's on the bottom.
I did note the gate of Q2 is more like 5.6v. I am not bright enough to know what to change around here to bring that up. All other voltages look good, except I cant get a good reading on the LU, I have it all bolten down and tefloned tubed over.
On my test speaker, it "Sounds" promising.
BTW, all caps are ELNA Silimic II's, except C1, which is a BHC slit foil. No C8,10 bypasses installed yet.
I will try to hook up the other channel this weekend, see if there is a large difference in voltages.
Thanks for the suggestions.
If you have close to a 48v rail, then drain of the Q3 should be about 42v, and half of this is your midpoint of about 21v (drain of Q2) which is approx your voltage on O/P cap C1.
So, with a steady current (bo signal) thru the amp of about 1,7Amps, the drain of Q2 should show this 21v (not 24) and its source should be about 4 volts = Q1 drain (direct connection) so measure at Q2 source.
The source of Q1 can be measured at the "top" of R3,4,5 and should be about 1 volt.
Power approx is from the top of the cct ...
Q5 is 1.7A x 5v = 8W,
Q3 is 1.7 x 21 = 36W,
R6,7,8 is 1.7 x 0.4 = 1W,
Q2 = 1.7 x 17v = 29W,
Q1 is 1.7 x 2.5v = 4W and
R3,4,5 is 1.7A x 1v = 2W
[total about 80W]
It looks like you have the Q's off board via wires - make sure the gate resistors are right on the gate, not left on pcb.
Also, see if you can get a really good C9 (approx 33pF) Some people leave it off altogether without any drama - At present I've got a 47pF styrene in series with a 68R resistor there and opens out to top a bit with better control of treble like cymbols, chimes and bell strikes, etc - got a bit fussy here - curious as to what would happen.
You can plug your h.phones directly to this amp and really have a close listen - works like a charm.
drcg,
Just take the parts off the R0 cct diagram - simple enough. The only real difference is folks like me suggesting fancy bits, but it's such a good result, suggest you spend the extra - if you can find some 99.99% silver wire (4 nines they call it!), that too! [1mm is about U$10/ft]
The Silmic caps are pretty standard, the O/P cap C1 can be confusing - the Jensen 4 poles aren't cheap and have cheaper version called F&T from Matt Campbell at "Audio Cap Co" or as TB, the Slit Foils (very clean mids) If you're flush, try some Sikorels.
There ARE slightly different sound M. Film power resistors - Recommended is the Panasonics - we have Phillips and Roedesteins here and are a slight bit softer - suggest NOT use the old wire wound, but many are used and seem okay - bit rough on top, is all.
Hope this helps ...
So, with a steady current (bo signal) thru the amp of about 1,7Amps, the drain of Q2 should show this 21v (not 24) and its source should be about 4 volts = Q1 drain (direct connection) so measure at Q2 source.
The source of Q1 can be measured at the "top" of R3,4,5 and should be about 1 volt.
Power approx is from the top of the cct ...
Q5 is 1.7A x 5v = 8W,
Q3 is 1.7 x 21 = 36W,
R6,7,8 is 1.7 x 0.4 = 1W,
Q2 = 1.7 x 17v = 29W,
Q1 is 1.7 x 2.5v = 4W and
R3,4,5 is 1.7A x 1v = 2W
[total about 80W]
It looks like you have the Q's off board via wires - make sure the gate resistors are right on the gate, not left on pcb.
Also, see if you can get a really good C9 (approx 33pF) Some people leave it off altogether without any drama - At present I've got a 47pF styrene in series with a 68R resistor there and opens out to top a bit with better control of treble like cymbols, chimes and bell strikes, etc - got a bit fussy here - curious as to what would happen.
You can plug your h.phones directly to this amp and really have a close listen - works like a charm.
drcg,
Just take the parts off the R0 cct diagram - simple enough. The only real difference is folks like me suggesting fancy bits, but it's such a good result, suggest you spend the extra - if you can find some 99.99% silver wire (4 nines they call it!), that too! [1mm is about U$10/ft]
The Silmic caps are pretty standard, the O/P cap C1 can be confusing - the Jensen 4 poles aren't cheap and have cheaper version called F&T from Matt Campbell at "Audio Cap Co" or as TB, the Slit Foils (very clean mids) If you're flush, try some Sikorels.
There ARE slightly different sound M. Film power resistors - Recommended is the Panasonics - we have Phillips and Roedesteins here and are a slight bit softer - suggest NOT use the old wire wound, but many are used and seem okay - bit rough on top, is all.
Hope this helps ...
Hi FFFo,
Sorry, small bridge wont take it, need bigger one - seriously suggest using 4 Shottky diodes per channel on a seperate h/sink to main amp body - the cooler, the better.
Peter Daniel used smaller caps on O/P to act as hipass filters - in fact I think he used 3 of them for 3 way speakers - ask him, email link on this site ...
If you're doing this, suggest Mundorf, Jensen, F&T, etc for base amp, BHC for the mids, and BHC, Rifa, PETPs, etc caps for treble - you might even find some Siemens Sikorels there somewhere!
Suggest adding a B1 buffer to input (and a good regulator off the amp's rails) to bring the inputZ up to user friendly size.
Sorry, small bridge wont take it, need bigger one - seriously suggest using 4 Shottky diodes per channel on a seperate h/sink to main amp body - the cooler, the better.
Peter Daniel used smaller caps on O/P to act as hipass filters - in fact I think he used 3 of them for 3 way speakers - ask him, email link on this site ...
If you're doing this, suggest Mundorf, Jensen, F&T, etc for base amp, BHC for the mids, and BHC, Rifa, PETPs, etc caps for treble - you might even find some Siemens Sikorels there somewhere!
Suggest adding a B1 buffer to input (and a good regulator off the amp's rails) to bring the inputZ up to user friendly size.
Hi VladimirK & James,
Thanks for the reply.
I will just like to understand whats the min volt and current required and why. I just dont understand why will a 25VDC 8A circuit requires up to 200V 35A rectifier, to me its an overkill, isnt it? Can you help to enlighten me?
Meanwhile, most available Shottky diodes are of low voltage and current. By the way, ultrafast diodes same as shottky?
Will a 68A 600V IXYS FRED Bridge Rectifier be good? or this On-Semi SCHOTTKY, 40A, 250V: http://sg.farnell.com/on-semiconductor/mbr40250tg/diode-schottky-40a-250v-to-220/dp/1431066
or this IXYS SCHOTTKY, 45A, 200V: http://sg.farnell.com/ixys-semiconductor/dsa90c200hb/diode-schottky-to-247ad/dp/1572508
Saw quite a few ppl mentioned about MUR860,600V 8A, is it enough for this application too?
" Suggest adding a B1 buffer to input to bring the inputZ up to user friendly size. "
This is a great idea.
" (and a good regulator off the amp's rails) ", what you suggest?
Oh yes, replacing those CRC supply with CLC, do you foresee the Inductor/Choke or <0.2ohm burning up?
Thanks for the reply.
I will just like to understand whats the min volt and current required and why. I just dont understand why will a 25VDC 8A circuit requires up to 200V 35A rectifier, to me its an overkill, isnt it? Can you help to enlighten me?
Meanwhile, most available Shottky diodes are of low voltage and current. By the way, ultrafast diodes same as shottky?
Will a 68A 600V IXYS FRED Bridge Rectifier be good? or this On-Semi SCHOTTKY, 40A, 250V: http://sg.farnell.com/on-semiconductor/mbr40250tg/diode-schottky-40a-250v-to-220/dp/1431066
or this IXYS SCHOTTKY, 45A, 200V: http://sg.farnell.com/ixys-semiconductor/dsa90c200hb/diode-schottky-to-247ad/dp/1572508
Saw quite a few ppl mentioned about MUR860,600V 8A, is it enough for this application too?
" Suggest adding a B1 buffer to input to bring the inputZ up to user friendly size. "
This is a great idea.
" (and a good regulator off the amp's rails) ", what you suggest?
Oh yes, replacing those CRC supply with CLC, do you foresee the Inductor/Choke or <0.2ohm burning up?
Hmmm, I didn't think about trying those FRED Bridges so don't know - not cheap, but convenient.
I've used MUR1610s, RURP820, MUR410, RURP860, all okay.
I found the BYW29s have better detail than above, and a slightly softer sound than the HFA08TB60 and ETH08s when used Silver wire inside.
I thought the Stealth ISL9R8120 had the same sound as the RURP820 - a little bit dull. (being picky, but easy to hear differences on h.phones!)
Look up the story about diode loading and peak reverse currents, etc.
Even with the current limiting Th in the primary path, the initial surge is a big one plus the latge sec caps, etc. Steady 2ndary current on this amp is less than 2 Amps/ ch.
I think minimum diodes would be about 80 volt, 10A - perhaps others can add to this?
There is a B1 pcb design coming together rather nicely in the Group Buy section complete with Salas shunt reg
- it's set up for +/- rails but easy to just make it with a single +ve rail off the F3's 42volt (after Q5).
- can do another more basic single supply and add a Lightspeed vol control in front of the buffer - this increases the "dynamic" sound of the amp quite a bit.
A lot of this U-beaut stuff depends on the speakers you use, the volume and the music content - opinions vary a lot - the "grain of salt" thing!
Choke supply - lot of people have used them okay - total series resistance of 0.2 R should be okay - I tried 0.1R 4.7mH HUGE air choke and a 0.3R 5H iron one, but preferred "faster sound" of Resistors - personal choice - both pretty good - that salt again.
[I'm using R-C-R-C of 0.1R, 22'000uF (F&T), 0.1R, 33'000uf Siemens (only size)]
The amp does seem to like silver ICs and speaker wire, but not sure with sources like Computer dacs, Blue Ray things, etc - I use NOS dac with Phillips chip.
Hi Vlad - amazing where the F3 shows up! Tashkent - images of old silk road, looong histories, modern changes.
I've used MUR1610s, RURP820, MUR410, RURP860, all okay.
I found the BYW29s have better detail than above, and a slightly softer sound than the HFA08TB60 and ETH08s when used Silver wire inside.
I thought the Stealth ISL9R8120 had the same sound as the RURP820 - a little bit dull. (being picky, but easy to hear differences on h.phones!)
Look up the story about diode loading and peak reverse currents, etc.
Even with the current limiting Th in the primary path, the initial surge is a big one plus the latge sec caps, etc. Steady 2ndary current on this amp is less than 2 Amps/ ch.
I think minimum diodes would be about 80 volt, 10A - perhaps others can add to this?
There is a B1 pcb design coming together rather nicely in the Group Buy section complete with Salas shunt reg
- it's set up for +/- rails but easy to just make it with a single +ve rail off the F3's 42volt (after Q5).
- can do another more basic single supply and add a Lightspeed vol control in front of the buffer - this increases the "dynamic" sound of the amp quite a bit.
A lot of this U-beaut stuff depends on the speakers you use, the volume and the music content - opinions vary a lot - the "grain of salt" thing!
Choke supply - lot of people have used them okay - total series resistance of 0.2 R should be okay - I tried 0.1R 4.7mH HUGE air choke and a 0.3R 5H iron one, but preferred "faster sound" of Resistors - personal choice - both pretty good - that salt again.
[I'm using R-C-R-C of 0.1R, 22'000uF (F&T), 0.1R, 33'000uf Siemens (only size)]
The amp does seem to like silver ICs and speaker wire, but not sure with sources like Computer dacs, Blue Ray things, etc - I use NOS dac with Phillips chip.
Hi Vlad - amazing where the F3 shows up! Tashkent - images of old silk road, looong histories, modern changes.
fff0 said:
In brief, the rectifier bridge loaded by C-L-C or C-R-C filter at constant current consumption, experiences charging current spikes through the diodes.
I enclosed an example picture here (post #17):
http://www.diyaudio.com/forums/showthread.php?postid=1850901#post1850901
During these short-period spikes charging current can achieve 20 amps value, and diodes start to suffer an increased voltage drop on them. The more powerful diode, the less problem with voltage drop. Minimum voltage drop problem show Shottky diodes, althogh they also differ substantially, one must care datasheets.
Less voltage drop makes power supply more efficient, all the rest being equal.
Also there could be an effect of trafo operation in such a spike-like mode - it switches from the idling to almost short circuit state of operation at the beginning of every current spike, and these repeating transient processes still have not got enough attention of researchers.
Summing up, the more powerful power supply you use for the same circuit, the more powerful bottom end you have.
Yeah, we grew up the legends, histories, fables and travel stories of Central Asia - Tamarlane, Hissan Valley, Kara Kum, Uba Tube, the first Universities of Bukhara, Silk Road, etc, etc - it's curious that one of the original centres of civilization is now a centre for petroleum and all the politics involved!
jameshillj said:
Actually not so much of natural gas and almost no petrol. Other "legs" of economy are agriculture, cotton, copper, uranium, gold, automotive ... This year our futball team "Bunyodkor" has lost to "Adelaida United" ...
But we hope to take revenge next time, Louis Felipe Scolari ( "Chelsea" ex-coach) is training our team now.
Hope to be excused for offtop.
Maybe, eh? ex pommie coach and all!
With the F3, I can't get over just how much you can adjust the sound of this relatively simple cct without sophisticated supplies. Added some snubbers and things to get rid of some mains rubbish and diode noise.
I thought about a shunt supply but the consumption made it impractical - now looking at a recent posting from the UGS power supply thread that looks quite promising - has a strange variety of Cmultiplier on output, apparently from Charles Hanson, so well worth a look. [tried to add the cct - too big, apparently]
With the F3, I can't get over just how much you can adjust the sound of this relatively simple cct without sophisticated supplies. Added some snubbers and things to get rid of some mains rubbish and diode noise.
I thought about a shunt supply but the consumption made it impractical - now looking at a recent posting from the UGS power supply thread that looks quite promising - has a strange variety of Cmultiplier on output, apparently from Charles Hanson, so well worth a look. [tried to add the cct - too big, apparently]