Hello everybody,
i would like to present you my new OTL amplifier, realised with cathode follower:
As you see, it is a standard schematic based on the projects of Andrea Ciuffoli in Headwize. It has no electrolytic capacitors, output caps are metal film AC caps. Output impedance is calculated to be ~40 Ohm.
Supply is with the lovely GZ34 and two 15H chokes:
No electrolytic caps here ether. Rated for 120mA at 300V the PSU is absolutely clean. PSU Designer II calculates output noise of less than 200uV.
This is the working model:
After finishing the fine tuning it will go in two separate boxes, that will stack one on top of the other.That is the reason the tubes are horisontal.
Here are the simulated tube regimes:
Here are some measurements made with my EMU0404 with Sony CDP970 used as a generator.
First the CDP970 only:
It is quiet clean, except the noise under 1Khz, but it is something related to the test setup and not relevant for the THD calculations.
Then measuring the driver only, between C1 and R8:
1st: -72dB and 2nd: -81dB
Then the whole amp, no load:
Same as the previous one. It seams that CF is not making any distortion when not loaded.
Then amp with 510 Ohm load:
1st: -73dB and 2nd: -97dB - Even better!
And finally the amp with 60 Ohm load:
Here the picture is quiet bad, but it does not really concern me since i will use the amp with high impedance phones.
So , i would be really happy to hear your opinion and critics to my realisation and design. I would really like advices on how to reduce THD even further without changing the topology. Any comment is welcome!
i would like to present you my new OTL amplifier, realised with cathode follower:
As you see, it is a standard schematic based on the projects of Andrea Ciuffoli in Headwize. It has no electrolytic capacitors, output caps are metal film AC caps. Output impedance is calculated to be ~40 Ohm.
Supply is with the lovely GZ34 and two 15H chokes:
No electrolytic caps here ether. Rated for 120mA at 300V the PSU is absolutely clean. PSU Designer II calculates output noise of less than 200uV.
This is the working model:
After finishing the fine tuning it will go in two separate boxes, that will stack one on top of the other.That is the reason the tubes are horisontal.
Here are the simulated tube regimes:
Here are some measurements made with my EMU0404 with Sony CDP970 used as a generator.
First the CDP970 only:
It is quiet clean, except the noise under 1Khz, but it is something related to the test setup and not relevant for the THD calculations.
Then measuring the driver only, between C1 and R8:
1st: -72dB and 2nd: -81dB
Then the whole amp, no load:
Same as the previous one. It seams that CF is not making any distortion when not loaded.
Then amp with 510 Ohm load:
1st: -73dB and 2nd: -97dB - Even better!
And finally the amp with 60 Ohm load:
Here the picture is quiet bad, but it does not really concern me since i will use the amp with high impedance phones.
So , i would be really happy to hear your opinion and critics to my realisation and design. I would really like advices on how to reduce THD even further without changing the topology. Any comment is welcome!
Last edited:
So it appears that the predominant source of 2nd harmonic at light loading is the first stage. You can reduce that a bit with a few circuit tricks, but at -72dB, really why bother? That's already pretty darn good.
If you're trying to win a spec war, go higher on the plate load resistor and raise the B+ to the first stage accordingly.
If you're trying to win a spec war, go higher on the plate load resistor and raise the B+ to the first stage accordingly.
Well it is fine, but i prefer to get as most as possible out of it, since it is a headphone amplifier and every distortion will be more noticeable.
And actually today i did that. I just wanted to see what will hapend with 15K anode load on the driver. I got something like 3-4db lower distortion when measuring between driver and CF , but when i load the schematic with 510R and measure it on the output it is 6dB higher distortion than before. It is pretty much strange to me. Probably it has to do something with R7 wich is only 1K for this set-up. It is possible that current movements in the CF are influencing the driver more with the smaller R7.
I will try to design a Power Supply with 4 different outputs for each tube. Hope that works better.
PS. I`m thinking about a double 6S45P in the CF. It will bring output impedance to less than 10R!
And actually today i did that. I just wanted to see what will hapend with 15K anode load on the driver. I got something like 3-4db lower distortion when measuring between driver and CF , but when i load the schematic with 510R and measure it on the output it is 6dB higher distortion than before. It is pretty much strange to me. Probably it has to do something with R7 wich is only 1K for this set-up. It is possible that current movements in the CF are influencing the driver more with the smaller R7.
I will try to design a Power Supply with 4 different outputs for each tube. Hope that works better.
PS. I`m thinking about a double 6S45P in the CF. It will bring output impedance to less than 10R!
So the next step for me is quiet radical:
I would like to use a negative PS for the CF in order to achieve the following things:
1. I can put the attenuator between the Driver and the CF. Since the CF input impedance is ~1M it is better to be there then in front of the driver that has ~150K input impedance.
2. I can use low voltage caps in the output like the BlackGate NX 2200uF 6,3V. They will be polarized with the bias voltage ~3,5V - high enough!
3. I will get higher P suppression & less ripple/noise
The only trade-off that I see is that the PSU price will double. It will rely on 4 Hammond 193J chokes(10H/200mA/82R) for the CF and the two chokes from my previous design.
Other than that, complete separation between the driver and the CF PSU are planed. Also I have choose to use 6 triodes in total for each CF. That way I will get a nice 15R output impedance, 75mA current and 3,5V Bias. For 300R headphones, this should be perfect.
Please comment, give ideas, criticize!
An externally hosted image should be here but it was not working when we last tested it.
I would like to use a negative PS for the CF in order to achieve the following things:
1. I can put the attenuator between the Driver and the CF. Since the CF input impedance is ~1M it is better to be there then in front of the driver that has ~150K input impedance.
2. I can use low voltage caps in the output like the BlackGate NX 2200uF 6,3V. They will be polarized with the bias voltage ~3,5V - high enough!
3. I will get higher P suppression & less ripple/noise
The only trade-off that I see is that the PSU price will double. It will rely on 4 Hammond 193J chokes(10H/200mA/82R) for the CF and the two chokes from my previous design.
Other than that, complete separation between the driver and the CF PSU are planed. Also I have choose to use 6 triodes in total for each CF. That way I will get a nice 15R output impedance, 75mA current and 3,5V Bias. For 300R headphones, this should be perfect.
Please comment, give ideas, criticize!
Hi!
i c'ant open the link
An externally hosted image should be here but it was not working when we last tested it.
your design of upgrade Ciuffoli
its very interesting
claudiomas
https://picasaweb.google.com/118012539153067336931/SOGNOAMPLICASQUEOTLNOS?authuser=0&feat=directlink
Hi,
Thanks for the interest. I am already using this configuration for a couple of months, but with Siemens E88CC as a driver and a double matched trio JAN Phillips E88CC for the CF. Sounds really nice!
Here is it again:
Regards,
K.
Thanks for the interest. I am already using this configuration for a couple of months, but with Siemens E88CC as a driver and a double matched trio JAN Phillips E88CC for the CF. Sounds really nice!
Here is it again:
An externally hosted image should be here but it was not working when we last tested it.
Regards,
K.
Last edited:
Colombo egg !
if evry channel get its own driver tube the sound image radically increase
in space and detail ,
the last and definitive measurement i use its by real audiophiles comments.
also 6 triodes for CF ....like V12 F1 engine, perfection c'ant last long,
noise, thd,fragile, for me ,
whay not KT 88 in pseudotriode ?
my last ciuffoli version whit 2X 6S9P (6SL7) and 2X 6P6S (6V6) nos reflector 60'
sound great, but your upgrade by symmetrical HT should be the
NEC PLUS ULTRA, no more electrolytics caps !!!
cheers
claudiomas
if evry channel get its own driver tube the sound image radically increase
in space and detail ,
the last and definitive measurement i use its by real audiophiles comments.
also 6 triodes for CF ....like V12 F1 engine, perfection c'ant last long,
noise, thd,fragile, for me ,
whay not KT 88 in pseudotriode ?
my last ciuffoli version whit 2X 6S9P (6SL7) and 2X 6P6S (6V6) nos reflector 60'
sound great, but your upgrade by symmetrical HT should be the
NEC PLUS ULTRA, no more electrolytics caps !!!
cheers
claudiomas
Thanks for the comment claudiomas!
Main reason for having 6 parallel triodes is to reduce the output impedance:
1 x KT88 = 12 mA/V GM
6 x E88CC = 75 mA/V GM
And the best part about the paralleling is that it is for a Cathode Follower where many of the problems you describe are not present.
I completely agree about your comment on the driver. Two tubes will be better than one for sure. And something that is not on Noval base will be even better! But this is something for the future.
Can you explain more about this NEC PLUS ULTRA? I am always interested in new PSU solutions.
Main reason for having 6 parallel triodes is to reduce the output impedance:
1 x KT88 = 12 mA/V GM
6 x E88CC = 75 mA/V GM
And the best part about the paralleling is that it is for a Cathode Follower where many of the problems you describe are not present.
I completely agree about your comment on the driver. Two tubes will be better than one for sure. And something that is not on Noval base will be even better! But this is something for the future.
Can you explain more about this NEC PLUS ULTRA? I am always interested in new PSU solutions.
(NEC PLUS ULTRA) {LATIN}...Hercule write this after open the sea at Gibraltar, that means:
nothing else after, the ultimate limit, c'ant be better,
PERFECTION.
still we never stop , and i'm sure you can do better whit KT88 ifyou like by grid polarization, even if need MKT output caps.
(i do now nothing about maths...but..)
Warning the heater-Katode voltage ,
for power supply i have a big lot of news 2200uF and 3300uF 200Volts,
so i use bridge rectifier.
cheers
claudiomas
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.