Member
Joined 2009
Paid Member
Not sure if this is real but looking at the schematic, Cdom is wrapped around the two-device VAS. I've seen a lower noise floor in the Harry77 sims if you don't do this. I put Cdom around the 2nd device only so that it's being driven from both ends and not from the LTP.
Bigun said:
🙁 I didn't understand a single word of this! Not to worry! Could some of our helpers please help with parts selection so we can get this going, PLEASE!!
Terry
Come on guys can we get some input on the capacitors, we can hardly call this an extreme amp with stock capacitors.
I know we have had some talk about the input and feedback capacitors but we fell short on consensus.
Hugh,
I for one would be very interested in your opinion here I note in other threads you have a passion in this area. Dont let cost or the current board dimensions hold you back here. we are talking extreme...
I know we have had some talk about the input and feedback capacitors but we fell short on consensus.
Hugh,
I for one would be very interested in your opinion here I note in other threads you have a passion in this area. Dont let cost or the current board dimensions hold you back here. we are talking extreme...
ULD Extreme
doylep
I am using 4.7uF 200V Auricap capacitors in the input of my Class A amplifier. The drawback is that they are bulky and expensive (Soundlabs) but quite transparent in comparison with your usual polyprops from Jaycar etc.
SandyK
doylep
I am using 4.7uF 200V Auricap capacitors in the input of my Class A amplifier. The drawback is that they are bulky and expensive (Soundlabs) but quite transparent in comparison with your usual polyprops from Jaycar etc.
SandyK
Hmm, a single bad cap choice can ruin an amp.
Silver mica for Cdom cap. It is worth carefully calibrating this cap to be absolutely sure you have the correct value. The value in the circuit may not be exactly right, and there is considerable influence on the SQ.
I would use a non-polar 22uF 6.3VW NX Black Gate for the input coupler. I do not agree with 47uF; it's completely over the top. The ideal choice here would be a Mundorf Silver Oil, but at 22uF this would be huge beyond practicality. I use 0.47uF in this application, though Zin is far higher. The BG is just 5mm in diameter and works very well! Large size in this cap can be problematic for RFI.
C2 should be a polyprop or even polystyrene.
Rail caps should be low ESR Nichicon Muse, or even the much cheaper Hitano EXRs. Rail bypasses can be any inexpensive mylar or polyester, same for collector caps.
Zobel cap should be a metallised polypropylene.
Power supply electros can be any quality component, but I like the Nichicon Muse, which are available from Michael Percy in Nevada City, CA.
Power supply rectification could be improved on this amp. Philips make a range of ultra fast soft recovery, rated to 15A continuous, which are very good. These DO have a very positive impact on SQ. The best power supplies for SS amps use low DCR chokes around 15mH, but I would not expect people to take this seriously for reasons of cost.
Hope this helps,
Hugh
Silver mica for Cdom cap. It is worth carefully calibrating this cap to be absolutely sure you have the correct value. The value in the circuit may not be exactly right, and there is considerable influence on the SQ.
I would use a non-polar 22uF 6.3VW NX Black Gate for the input coupler. I do not agree with 47uF; it's completely over the top. The ideal choice here would be a Mundorf Silver Oil, but at 22uF this would be huge beyond practicality. I use 0.47uF in this application, though Zin is far higher. The BG is just 5mm in diameter and works very well! Large size in this cap can be problematic for RFI.
C2 should be a polyprop or even polystyrene.
Rail caps should be low ESR Nichicon Muse, or even the much cheaper Hitano EXRs. Rail bypasses can be any inexpensive mylar or polyester, same for collector caps.
Zobel cap should be a metallised polypropylene.
Power supply electros can be any quality component, but I like the Nichicon Muse, which are available from Michael Percy in Nevada City, CA.
Power supply rectification could be improved on this amp. Philips make a range of ultra fast soft recovery, rated to 15A continuous, which are very good. These DO have a very positive impact on SQ. The best power supplies for SS amps use low DCR chokes around 15mH, but I would not expect people to take this seriously for reasons of cost.
Hope this helps,
Hugh
Sorry for neglecting this thread and not responding sooner. My day job has buried me of late. Argh ...
The PCB was laid out using stocked parts from Mouser and Digikey. Many of the pads (especially for caps) were laid out for multiple vendor and package sizes. However a couple of part layouts are specific. Those are ...
C1 = Nichicon MUSE 47uF @ 16 volts (Mouser)
C2 = CDE Silver Mica 870 pF
R27 thru 32 = XICON Vertical cement .22 Ohm @ 5 watts (Mouser)
C12 = VISHAY BFC237351154 150nF @ 400 volts (Digikey)
The resistors are all XICON metal films (Mouser)
The electrolytic caps are all Panasonic FCs (Digikey)
The rectangular box caps are all WIMAs (Mouser)
For the output inductor I was planning to strip some #14 magnet wire from a loudspeaker inductor that I have loitering in my parts bin. Parts Express is a good source for those lacking a parts stash. I had simply planned on putting wraps on a convenient diameter (3/8th of an inch??) and pulling off coils until I measured the target 6.8uH value.
The PCB was laid out using stocked parts from Mouser and Digikey. Many of the pads (especially for caps) were laid out for multiple vendor and package sizes. However a couple of part layouts are specific. Those are ...
C1 = Nichicon MUSE 47uF @ 16 volts (Mouser)
C2 = CDE Silver Mica 870 pF
R27 thru 32 = XICON Vertical cement .22 Ohm @ 5 watts (Mouser)
C12 = VISHAY BFC237351154 150nF @ 400 volts (Digikey)
The resistors are all XICON metal films (Mouser)
The electrolytic caps are all Panasonic FCs (Digikey)
The rectangular box caps are all WIMAs (Mouser)
For the output inductor I was planning to strip some #14 magnet wire from a loudspeaker inductor that I have loitering in my parts bin. Parts Express is a good source for those lacking a parts stash. I had simply planned on putting wraps on a convenient diameter (3/8th of an inch??) and pulling off coils until I measured the target 6.8uH value.
Bigun said:
In this design which is a variation on the DS blameless theme the 2 VAS transistors are in effect one high beta device so it's correct that both are nested within cdom. This results in a much lower VAS output impedance (XC/beta) which in turn reduces some of the gm doubling induced distortions.
Carl,
Just a question from the not so informed. Parts Express has a 6.8 MH coil inductor. Is that the same as 6.8 UH. New territory here for me. Also, 50.00 plus dollars. Ouch.
Tad
Just a question from the not so informed. Parts Express has a 6.8 MH coil inductor. Is that the same as 6.8 UH. New territory here for me. Also, 50.00 plus dollars. Ouch.
Tad
6.8mH is one thousand times more inductance than 6.8uH
For 6.8uH all you need is a short length of enameled copper wire.
Choose wire between 0.5mm and 3mm diameter to suit your desired maximum resistance.
Many now suggest that 6.8uH is far too high for this type of amplifier.
Try 3uH or even as low as 1uH. Stability may be just as good with any of these values.
Bypass the inductor with a parallel resistor <=speaker impedance. Try going as low as half speaker impedance.
M= Mega 10^6
m=milli 10^-3
U=?
u=micro 10^-6
For 6.8uH all you need is a short length of enameled copper wire.
Choose wire between 0.5mm and 3mm diameter to suit your desired maximum resistance.
Many now suggest that 6.8uH is far too high for this type of amplifier.
Try 3uH or even as low as 1uH. Stability may be just as good with any of these values.
Bypass the inductor with a parallel resistor <=speaker impedance. Try going as low as half speaker impedance.
M= Mega 10^6
m=milli 10^-3
U=?
u=micro 10^-6
proto update
having difficulties sourcing toshiba 2SB649A/2SD669A (probably due to the fact they dont make em anymore: PMC do though), could not find any info in the thread as to why these were chosen over the original BF 470/469 Does anyone have any information on this? are we ok to use the original devices ?
ps. The boards have left customs...
having difficulties sourcing toshiba 2SB649A/2SD669A (probably due to the fact they dont make em anymore: PMC do though), could not find any info in the thread as to why these were chosen over the original BF 470/469 Does anyone have any information on this? are we ok to use the original devices ?
ps. The boards have left customs...
Capacitor Selection
Thanks guys for your input on the caps,
I will put together a list of components soon for your review.
ps. Carl, I know what you mean work does get in the way some times; I will be sidetracked too this week...
Thanks guys for your input on the caps,
I will put together a list of components soon for your review.
ps. Carl, I know what you mean work does get in the way some times; I will be sidetracked too this week...
Carl et al,
To make a point brought up earlier in this thread regarding the vbe and diode string I was wondering if the approach used by Mcintosh in there thermal trak amps is as bad as I have read here.
To obtain a relatively "perfect" equalization in the vbe Mac chose to incorporate a correction ic in the amp. How this is done is beyond me it just seems practical with the huge choice in small chip devices available. If the preferred target can be figured and controlled within a few milliamps then is that not the the desired effect how ever it is done. With the large heat range from startup to extended play being uncontrollable then constantly varying vbe to compensate makes good sense if perfection is what diy is all about.
Tad
To make a point brought up earlier in this thread regarding the vbe and diode string I was wondering if the approach used by Mcintosh in there thermal trak amps is as bad as I have read here.
To obtain a relatively "perfect" equalization in the vbe Mac chose to incorporate a correction ic in the amp. How this is done is beyond me it just seems practical with the huge choice in small chip devices available. If the preferred target can be figured and controlled within a few milliamps then is that not the the desired effect how ever it is done. With the large heat range from startup to extended play being uncontrollable then constantly varying vbe to compensate makes good sense if perfection is what diy is all about.
Tad
Capacitor Selection
Hugh,
differences to your list are a silver mica in C2 and could only fit a metallised polyester in C12. What so you think of these choices? Do you have any sugestions for replacing Q7 2SB649A and Q9 2SD669A as they are not readily available.
C1 47uF NP percy Audio Black Gate 22ìF . . .6.3V NX HiQ . . . . . .2.25 ..............5mm x 7mm
C2 820pF rs silver mica
C3 47uF 35V percy Audio Nichicon 47ìF ...................50V KZ ........... .75...........10mm x 16mm
C4 47uF 35V percy Audio Nichicon 47ìF ...................50V KZ ........... .75...........10mm x 16mm
C5 220uF 16V percy Audio Nichicon 220ìF .................16V FG ........... .70........10mm x 12.5mm
C6 100pF 100V rs silver mica
C7 470uF 63V percy Audio Nichicon 470ìF ...............100V FG ......... 2.95........18mm x 35.5mm
C8 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C9 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C9a-9j 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C10 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C11 1uF rs evox RIFA Metallised Polyester 63Vdc MMK5 105K63J04L4
C12 150nF 400V rs VISHAY MKT1822-415/405 Metallised Polyester 400V
C13 1000uF 63V percy Audio Nichicon 1000ìF ............100V FW ......... 2.25...........18mm x 40mm
C14 100nF (10mm) rs EPCOS Metallised polyester B32521C3104J 250V
C15 1000uF 63V percy Audio Nichicon 1000ìF ............100V FW ......... 2.25...........18mm x 40mm
C16 100nF (10mm) rs EPCOS Metallised polyester B32521C3104J 250V
Hugh,
differences to your list are a silver mica in C2 and could only fit a metallised polyester in C12. What so you think of these choices? Do you have any sugestions for replacing Q7 2SB649A and Q9 2SD669A as they are not readily available.
C1 47uF NP percy Audio Black Gate 22ìF . . .6.3V NX HiQ . . . . . .2.25 ..............5mm x 7mm
C2 820pF rs silver mica
C3 47uF 35V percy Audio Nichicon 47ìF ...................50V KZ ........... .75...........10mm x 16mm
C4 47uF 35V percy Audio Nichicon 47ìF ...................50V KZ ........... .75...........10mm x 16mm
C5 220uF 16V percy Audio Nichicon 220ìF .................16V FG ........... .70........10mm x 12.5mm
C6 100pF 100V rs silver mica
C7 470uF 63V percy Audio Nichicon 470ìF ...............100V FG ......... 2.95........18mm x 35.5mm
C8 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C9 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C9a-9j 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C10 100nF rs AVX metallised Polyester, 250V BQ074G0104K
C11 1uF rs evox RIFA Metallised Polyester 63Vdc MMK5 105K63J04L4
C12 150nF 400V rs VISHAY MKT1822-415/405 Metallised Polyester 400V
C13 1000uF 63V percy Audio Nichicon 1000ìF ............100V FW ......... 2.25...........18mm x 40mm
C14 100nF (10mm) rs EPCOS Metallised polyester B32521C3104J 250V
C15 1000uF 63V percy Audio Nichicon 1000ìF ............100V FW ......... 2.25...........18mm x 40mm
C16 100nF (10mm) rs EPCOS Metallised polyester B32521C3104J 250V
Peter,
My schemat is V2.0i, with no designations, so I will assume that C2 is the LPF in the input - silver mica is fine and dandy and my preference anyway - and C12 is the 150nF Zobel cap on the output. For this cap, which takes the full output swing of the amp and copes with reactive kickback from the speakers, polyprop is preferred because it is, to some degree, self-healing. However, an MKS is just fine, particularly at that very high rating.
The CCS device, Q7 (pnp) and VAS Q9 (npn), could be BF470, a Philips 250V, low Cob 70MHz device and 2SC3423, a 150V 1.8pF Cob 50mA 200MHz device from Toshiba, respectively. I use them exclusively and they are wonderful devices. It is helpful to have the CCS a tad slower than the VAS.
Hope this helps,
Hugh
Hugh,
Thanks for that, I think there are some shortcommings with this layout we may have gone a little too compact. I have been sifting though this thread and found other reccomendations from yourself as follows
Q7 2SA1360
Q8 2SC1845 Reduce emitter resistor from 2K2 to 470R.
Q9 2SC3423
Do these mods still stand ?
ps. I have attached latest schematic,
the prototype boards have finally arrived...
Thanks for that, I think there are some shortcommings with this layout we may have gone a little too compact. I have been sifting though this thread and found other reccomendations from yourself as follows
Q7 2SA1360
Q8 2SC1845 Reduce emitter resistor from 2K2 to 470R.
Q9 2SC3423
Do these mods still stand ?
ps. I have attached latest schematic,
the prototype boards have finally arrived...
Attachments
ULD Extreme
doylep
I believe that the 2K2 emitter resistor of Q8 should stay as is.
You could always experiment afterwards.
SandyK
doylep
I believe that the 2K2 emitter resistor of Q8 should stay as is.
You could always experiment afterwards.
SandyK
Hello Doyle and Hugh,
The current schematic with all the changes and part numbers that matches the PCB layout is on Dudainc's website that is referenced back in Post #260.
Nothing has changed since those files were posted. The PCB was laid out assuming parts readily available from Mouser and Digikey. The PCB is an evolution of a version that I successfully built some time ago. There were no parts placement issues when building that version of the amplifier. But then again those were common parts from Mouser and Digikey.
Nichicon electrolytics are physically larger than the Panansonic FC caps that were assumed when the PCB was laid out. I had assumed only one Nichicon cap (C5) when the board was laid out.
The current schematic with all the changes and part numbers that matches the PCB layout is on Dudainc's website that is referenced back in Post #260.
Nothing has changed since those files were posted. The PCB was laid out assuming parts readily available from Mouser and Digikey. The PCB is an evolution of a version that I successfully built some time ago. There were no parts placement issues when building that version of the amplifier. But then again those were common parts from Mouser and Digikey.
Nichicon electrolytics are physically larger than the Panansonic FC caps that were assumed when the PCB was laid out. I had assumed only one Nichicon cap (C5) when the board was laid out.
Peter,
Stay with 2K2 rather than 470R for the EF load resistor...... I agree with Sandy.
The A1360 is a Toshiba device, complement to C3423 in fact, and while it is fast, I always prefer to have the active device, the C3423, faster than the CCS which serves it. So leave the BF470 as is, bearing in mind it has very low Cob anyway, around 2.5pF.
Carl,
I think your layout is brilliant, and I doffs me cap. I certainly could do no better. I also applaud your component choices, Digikey and Mouser are excellent references, and your choices should make the amp rather better than the original. I've a couple of the SC boards here, courtesy Jeremy Howard, and they are NOT well designed at all.
My concern about this amp rests with the biasing scheme and the new diode equipped outputs. There is sufficient flexiibility in Carl's design to skirt around this should it prove tetchy, however, and I strongly endorse Carl's use of higher emitter resistors, 0.22R rather than 0.1R originally specified.
Optimal output stage bias changes with emitter resistor choice. It should be 26mV regardless of this resistor value; for 0.47R this means 55mA, for 0.22R it means 118mA, close to that actually recommended in the original design. For 0.1R, note, it means the optimal bias level in the output stage should be 260mA!!
Cheers,
Hugh
Stay with 2K2 rather than 470R for the EF load resistor...... I agree with Sandy.
The A1360 is a Toshiba device, complement to C3423 in fact, and while it is fast, I always prefer to have the active device, the C3423, faster than the CCS which serves it. So leave the BF470 as is, bearing in mind it has very low Cob anyway, around 2.5pF.
Carl,
I think your layout is brilliant, and I doffs me cap. I certainly could do no better. I also applaud your component choices, Digikey and Mouser are excellent references, and your choices should make the amp rather better than the original. I've a couple of the SC boards here, courtesy Jeremy Howard, and they are NOT well designed at all.
My concern about this amp rests with the biasing scheme and the new diode equipped outputs. There is sufficient flexiibility in Carl's design to skirt around this should it prove tetchy, however, and I strongly endorse Carl's use of higher emitter resistors, 0.22R rather than 0.1R originally specified.
Optimal output stage bias changes with emitter resistor choice. It should be 26mV regardless of this resistor value; for 0.47R this means 55mA, for 0.22R it means 118mA, close to that actually recommended in the original design. For 0.1R, note, it means the optimal bias level in the output stage should be 260mA!!
Cheers,
Hugh
Thank you for the kind words Hugh,
The emitter resistors in the final schematic are shown as .22R. Those are Mouser #280-PRM5-0.22-RC. The PCB layout assumes that part number. The value can change but the vendor and part family cannot without placement problems.
You can use a Nichicon MUSE SE 47uF 16 volt cap for C1 if you like. That one fits.
The emitter resistors in the final schematic are shown as .22R. Those are Mouser #280-PRM5-0.22-RC. The PCB layout assumes that part number. The value can change but the vendor and part family cannot without placement problems.
You can use a Nichicon MUSE SE 47uF 16 volt cap for C1 if you like. That one fits.
- Status
- Not open for further replies.