Eli Duttman said:
If all he's got is 30W, then he doesn't need to go Class AB2 at all. Look at the spec sheet:
Vpp= 365Vdc
Vsgsg= 270Vdc
Vgk= -22.5Vdc
Ipq= 48mA
Rl= 6K6 (P-2-P)
Po= 26.5W
THD= 1.8%
Right there, you have a really good operating point for both the 807 and 6L6. Add that local NFB to help clean up the harmonic mess and you have an excellent sounding amp. You aren't gonna do any better at the 30W limit. Add a decent grid driver (I used 6SN7 cathode followers) and screen voltage active regulation (solid state or hollow state -- doesn't matter) and you can push it just a bit into Class AB2 to get a few extra watts.
That's what I used: 30W Hammond OPTs, and I get core saturation before actual clipping when it goes to 11.
chrish said:
I just derive mine from the main power xfmr. In one design, this was a solid state +/- supply. Since the current draw from B- is so much smaller, an RC filter works out very well, and the filter resistor can also be used to drop unwanted volts.
In another design, the positive rail was based on the 5U4GB hollow state power diode, and the negative rail derived with silicon diodes. This worked out well since the power xfmr would over volt with SS rectification, but the forward drop of the 5U4GB and voltage averaging at the first resevoir capacitor got the Vpp right where I wanted it. A SS negative rail also works out just great in that the bias is already there before the heaters warm up, so nothing takes a high voltage hit.
Miles' comments have prompted me to carefully look at the GE 6L6GC data sheet.
GE shows a pentode mode AB2 condition set that yields 31 W./2% THD. The set includes: a 6 KOhm A-A load, 360 V. on the plate, 225 V. on g2, and a 78 mA. "idle" IB. This project will use ultralinear mode "finals", for distortion and damping reasons. A 400 V. B+ rail, 100 Ohm g2 stopper/limiters and a 70 mA. "idle" cathode current appear to be a reasonable combination.
I firmly believe that infrasonic trash should be rolled off at the I/P of a circuit that includes the O/P trafo in a NFB loop. Mullard style circuitry is such a case. Examination of the spec's for the Tamura trafos to be used shows them to be fully rated down to 30 Hz. With that in mind, I suggest a 68 nF. cap./120 KOhm grid leak combination. F3 for that paring is 19.5 Hz. That appears to be a reasonable compromise between bandwidth and O/P trafo core saturation protection.
GE shows a pentode mode AB2 condition set that yields 31 W./2% THD. The set includes: a 6 KOhm A-A load, 360 V. on the plate, 225 V. on g2, and a 78 mA. "idle" IB. This project will use ultralinear mode "finals", for distortion and damping reasons. A 400 V. B+ rail, 100 Ohm g2 stopper/limiters and a 70 mA. "idle" cathode current appear to be a reasonable combination.
I firmly believe that infrasonic trash should be rolled off at the I/P of a circuit that includes the O/P trafo in a NFB loop. Mullard style circuitry is such a case. Examination of the spec's for the Tamura trafos to be used shows them to be fully rated down to 30 Hz. With that in mind, I suggest a 68 nF. cap./120 KOhm grid leak combination. F3 for that paring is 19.5 Hz. That appears to be a reasonable compromise between bandwidth and O/P trafo core saturation protection.
That's what I did for my KT88 amp. I did simple regulators so that I don't have to re-work the power supply if I make changes and current draws change. You're welcome to adapt the relevant parts if you wish.
My bias regulation is absolute since so is my B+. One could easily substitute a resistor for the 100v zener to make bias voltage changes counteract B+ voltage changes.
Attachments
Eli Duttman said:
Not wanting to go too far off topic, but couldn't the global feedback loop be taken from the primary of the OPT in order to get more phase margin, relax the bandwidth requirements on the OPT, or both? Crowhurst wrote about this and Dennis Collin published an article comparing feedback effects from primary and secondary. His article was in AudioXpress, and I believe was a preliminary work to his Mad Katy amp.
weinstro said:
Rob,
The Tamura O/P trafos Chris is going to use provide a dedicated tertiary NFB winding. Taking OEM "hints" seriously is (IMO) sound policy. For a "classic" design that uses such "iron", look at the Marantz 8B.
BTW, the lowest fundamental a double bass can produce is approx. 31 Hz. IMO, a F3 of 19.5 Hz. presents little, if any, problem in the real music dept. Remember, the high pass pole is outside of the global NFB loop.
Joining in very late, primarily to say, what a delight to see some proper specs again on output transformers!
About halfway through this I was also wondering why all the 'special' shenannigans for 30W? 6L6GC easily gives 40W in UL with about 500V B+ and cathode bias, class AB-1.
As an aside, Crish, could you give an idea about the price of your particular 30W OPT?
About halfway through this I was also wondering why all the 'special' shenannigans for 30W? 6L6GC easily gives 40W in UL with about 500V B+ and cathode bias, class AB-1.
As an aside, Crish, could you give an idea about the price of your particular 30W OPT?
Oh yes, sorry - I also wanted to say about frequency response:
Looking at hearing response graphs quite some energy would be needed to make 30 Hz audible with any ordinary loudspeaker. I doubt that anything lower could be audible with only 30W from the amplifier, depending. The threshold of hearing rises quite sharply at that frequency.
Looking at hearing response graphs quite some energy would be needed to make 30 Hz audible with any ordinary loudspeaker. I doubt that anything lower could be audible with only 30W from the amplifier, depending. The threshold of hearing rises quite sharply at that frequency.
Over on AA, Doug Piccard (Bandersnatch) asked for info. about the AcroSound TO-350 O/P trafo. IMO, that "iron" would stone cold KILL, when mated to 807s.
Just look at this set of conditions obtained from TDSL.
Class Va Vg2 Vg1 Ia Ig2 Ra S Rk Zout Pout THD
AB2 P/P 600 300 -30.0 30-100 2.5-10.5 6,400 80.0 3.5
Regulated 300 VDC applied to the tertiary winding allows for full bore, maxed out, performance in UL mode. It seems that not much in the way of GNFB would be needed, to obtain both low distortion and good damping.
Just look at this set of conditions obtained from TDSL.
Class Va Vg2 Vg1 Ia Ig2 Ra S Rk Zout Pout THD
AB2 P/P 600 300 -30.0 30-100 2.5-10.5 6,400 80.0 3.5
Regulated 300 VDC applied to the tertiary winding allows for full bore, maxed out, performance in UL mode. It seems that not much in the way of GNFB would be needed, to obtain both low distortion and good damping.
Thanks for all of the great contributions and thought that you have all put in to this thread, I really do appreciate it.
My situation at the moment is that as i have been busy with work/moving/university, my partner would be a bit annoyed if I dove in to this project to the exclusion of her... In other words, I have to work on it slowly and use time I have while away from home with work for the research rather than time at home. This is why I have been a little quiet on the subject in the last few weeks.
Anyway, I travel a bit for work, mainly to Asia, but I will be travelling to L.A. in mid January, so was thinking of ordering the power iron for this project and picking it up while I am there. I think that Eli was suggesting a B+ of around 400 volts. Others were suggesting around 500 volts. For a 400 volt B+ I was thinking perhaps of the Hammond 372JX 300-0-300 250mA, 5VCT 4A, 6.3VCT 8A. For higher voltage I would have to go for something like a Hammond 378X 400-0-400 200mA.
Design changes and evolution are not a problem, and i am grateful for the help so far, but I think I need to get a firm idea of power requirement soon so I can organise the power transformer.
So are we still thinking 12AT7 voltage amp, DC coupled to an ECC99 LTP, mosfet source follower DC coupled to 6L6 in UL with some local NFB from the dedicated winding on the Tamura transformer? I have several 12AT7/ECC81. I have a few 6N6P valves in the box. They are described as an ECC99 equivalent (but with 6.3 volt heater). Would they be suitable to use in the LTP?
Thanks again for the interest and help.
Regards,
Chris
My situation at the moment is that as i have been busy with work/moving/university, my partner would be a bit annoyed if I dove in to this project to the exclusion of her... In other words, I have to work on it slowly and use time I have while away from home with work for the research rather than time at home. This is why I have been a little quiet on the subject in the last few weeks.
Anyway, I travel a bit for work, mainly to Asia, but I will be travelling to L.A. in mid January, so was thinking of ordering the power iron for this project and picking it up while I am there. I think that Eli was suggesting a B+ of around 400 volts. Others were suggesting around 500 volts. For a 400 volt B+ I was thinking perhaps of the Hammond 372JX 300-0-300 250mA, 5VCT 4A, 6.3VCT 8A. For higher voltage I would have to go for something like a Hammond 378X 400-0-400 200mA.
Design changes and evolution are not a problem, and i am grateful for the help so far, but I think I need to get a firm idea of power requirement soon so I can organise the power transformer.
So are we still thinking 12AT7 voltage amp, DC coupled to an ECC99 LTP, mosfet source follower DC coupled to 6L6 in UL with some local NFB from the dedicated winding on the Tamura transformer? I have several 12AT7/ECC81. I have a few 6N6P valves in the box. They are described as an ECC99 equivalent (but with 6.3 volt heater). Would they be suitable to use in the LTP?
Thanks again for the interest and help.
Regards,
Chris
chrish said:
Chris,
It sounds like you are well on your way to a good plan. I am with you completely about the use of a source follower to drive a 6L6 into AB2. That's definitely the way to go for the most power while still using a minimalist approach.
I would offer a couple of suggestions about nfb and the input/driver stage though. Unless the input/phase splitter/driver stage is implemented poorly 95 percent of distortion, of whatever kind you are talking about, will come from the final amplification stage. Considering this fact it might be wise to consider feedback from the OPT to the 6L6 through their cathodes. Just ignore gfb. The only requirement would be that you get an OPT with a center tapped secondary. They are defintely made and shouldn't be too hard or expensive to find. Electraprint probably makes them. You'll have much better phase margin and with AB2 you'll really be able to crank on the power without getting into high distortion. You don't need special windings for this and it might even be detrimental as feedback levels over and above speaker level feedback might cause an overly constricted sound.
The other thing I would suggest, especially since you seem to agree that minimalist design is best, which I tend to agree with - why not just a single input/phase splitter stage? A 12AX7 is a really good tube despite being popular and has a gain of 100. It should easily be able to drive the low current requirements of of a mosfet source follower from a standard 2 volt rms input. It would also have high enough output gain to drive the 6L6 into AB2 with enough drive to spare with local final stage cathode feedback. (Maybe not enough with GFB.) If you are worried about the rather thin current driving ability of the 12ax7 you use 2 of them in parallel. Just use the LTP pair topology with 2 12ax7s in parallel and a single CCS running twice the current. Because a 12ax7 is a dual tube you would need 2 tubes for each channel. You'll increase the signal to noise of the input stage significantly this way and virtually eliminate any worries about current driving ability if you do have them.
This topology should easily provide enough gain for cathode feedback in the final stage, provide a simpler circuit with less gain stages, and have better signal to noise.
Current-driving won't be an issue. ECC83/12AX7 love high plate loads, especially CCS. Let's look at the gain structure: in CCS, you're likely to get a gain of 45. It will take 36V peak to drive the output tube grids. If you wrap the cathodes around the secondary, that raises the value to roughly 60V, so you're probably in the ballpark. The caution will be that 12AX7 tends to start drawing its own grid current below a volt between cathode and grid and has a large Miller capacitance, so the input circuit and the driving preamp will have to be low impedance to keep bandwidth up and distortion down.
Chris,
The NFB winding on your Tamura trafos does not have much current handling capability. It is intended for use driving "textbook" GNFB loops. Do not use it for local NFB in the O/P stage. The advantage of the separate winding lies in decoupling the NFB circuitry from the speaker load. The Tamura data sheet shows both the speaker winding and the NFB winding as being grounded. However, the presence of the separate NFB winding allows you to float the speaker connections. Experimentation with both arrangements is probably in order.
The NFB winding on your Tamura trafos does not have much current handling capability. It is intended for use driving "textbook" GNFB loops. Do not use it for local NFB in the O/P stage. The advantage of the separate winding lies in decoupling the NFB circuitry from the speaker load. The Tamura data sheet shows both the speaker winding and the NFB winding as being grounded. However, the presence of the separate NFB winding allows you to float the speaker connections. Experimentation with both arrangements is probably in order.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.