Joe
I asked you another day, why you use such high resistor values for the feedback.
While reading this article http://www.du.edu/~etuttle/electron/elect3.htm
I remarked the feedback with the T resistor network, this way lowering the value of the feedback path.
Could this be a way, to improve your circuit (a little bit) and give more freedom to work with higher input-Z?
Franz
I asked you another day, why you use such high resistor values for the feedback.
While reading this article http://www.du.edu/~etuttle/electron/elect3.htm
I remarked the feedback with the T resistor network, this way lowering the value of the feedback path.
Could this be a way, to improve your circuit (a little bit) and give more freedom to work with higher input-Z?
Franz
Franz G said:
Yes Franz, it is an interesting technique, albeit it adds two more components.
With the current arrangement, this gives me a 22K7 input Z, that's quite workable and the desired LF roll-off I desired was gained by using 3u3uF input cap.
Yes, I did very carefully think about using the 1M f/b value to give gain = 44 or 32.9dB, but as far as DC stablility was concerned, it was near impeccable. Also, with the gain set so high there is effectively no f/b above 200KHz. This minimises capacitive effects/coupling along with some basic attention to layout. Add to this, the LPF curtails bandwidth effectively (also makes it sound better). In the end I was very happy with the results.
But I did do a "what if" - like what if I needed 47K input Z value. Using 10K x 2 and 47R to ground, this would give a gain very close, 46.6 or 33.3dB. So it is a good technique to keep in mind.
If I wanted 100K input, then use 22R for similar gain = 33.1dB
BTW, I did not bother with the rather intricate maths, much easier to set up values in CircuitMaker with 1V RMS in and an "RMS Meter" on the output.
Joe R.
fdegrove said:
This confirms what I have found myself, the Dutch Amperex and the very similar construction of the Mullards, are the best I have used, IMHO, in the JLTi. This is despite the fact that I supply them with the Sovtek "6922" Reflector [current] type. Has anyone a confirmation that the Amperex/Mullard version is now being made by someone else? I'd like to have a listen and ditch the Sovteks (provided the dollars are reasonable). They are not bad, but better is to be had.
Joe R.
Hi,
Joe, et all,
You heard someone else is making these?
I heard about some obscure Mexican/South-American plants making some valves for OEMs but that's about it...No idea what quality that would be, not even sure it's true.
If you want to ditch the Sovtek 6922s currently in production, the two remaining European manufacturers, Ei and JJ, are about all you can chose from, both will likely be more expensive to buy from than the Sovtek ones.
There are of course the Chinese ECC88/6DJ8 but my tests show them to be closer to a 6N1-P than a ECC88...
Not surprisingly as a lot of the Chinese parts used to build tubes are imported from the former USSR.
BTW, the historical stuff "narrated" in the link I provided is absolutely far from accurate. In fact most of it is simple false but I suppose the guy meant well....
The sound impressions correlate mostly with what I experienced myself, which is what matters most...To me that is.
Cheers, 😉
Joe, et all,
You heard someone else is making these?
I heard about some obscure Mexican/South-American plants making some valves for OEMs but that's about it...No idea what quality that would be, not even sure it's true.
If you want to ditch the Sovtek 6922s currently in production, the two remaining European manufacturers, Ei and JJ, are about all you can chose from, both will likely be more expensive to buy from than the Sovtek ones.
There are of course the Chinese ECC88/6DJ8 but my tests show them to be closer to a 6N1-P than a ECC88...
Not surprisingly as a lot of the Chinese parts used to build tubes are imported from the former USSR.
BTW, the historical stuff "narrated" in the link I provided is absolutely far from accurate. In fact most of it is simple false but I suppose the guy meant well....
The sound impressions correlate mostly with what I experienced myself, which is what matters most...To me that is.
Cheers, 😉
Joe Rasmussen said:
Gday Joe,
Agreed WRT the Russian 6922 Vs Amperex varieties.
Here's an option worth trying: Ask Steve G to Cryo
treat some Russian 6922's for you and re compare them to
your reference Amperexes.
This *supposedly* brings them closer at a fraction of the cost
although I haven't personally listened to them.
Cheers,
Terry
Terry_Demol said:
I sent off some of my 6H23pi for the cryo treatment (althou they get a lot more care than just a dunk in the nitrogen bath, Bill has a pretty detailed description in this ebay auction
Be interesting to hear them vrs stock when they return.
dave
T network for feedback
Joe
Is there a free version of this Circuit Maker available?
Here I offer an Excel sheet to calculate the t-network feedback path.
Just change the value of the resistors and you get as result the "equivalent" Rf and the gain.
Franz
Joe
Is there a free version of this Circuit Maker available?
Here I offer an Excel sheet to calculate the t-network feedback path.
Just change the value of the resistors and you get as result the "equivalent" Rf and the gain.
Franz
Terry_Demol said:
Hey Tez, good to see you bob here.
Got news for ya, got two Cryoed Amperexes right here that Steve dropped up and will plug them in tonite. Will let you know how it goes but my anticipation is high. If I like them Steve might find my front door closed. 😀
Interesting idea - got plenty of old and new style Sovteks here, get a pair if each cryoed - a must try.
Oh, BTW, got another level out of the 555 - even Steve and Dragon liked it... and a lot was my impression.
Joe
Re: T network for feedback
I don't think so, but I have been told it's available as warez. There is also available a 'student version' which is free but limited number of components - came across someone mentioning that. Try to Google it.
Joe R.
Franz G said:
I don't think so, but I have been told it's available as warez. There is also available a 'student version' which is free but limited number of components - came across someone mentioning that. Try to Google it.
Joe R.
Re: Re: T network for feedback
http://www.altium.com/circuitmaker/downloads/student.htm
Joe Rasmussen said:
http://www.altium.com/circuitmaker/downloads/student.htm
Thanks! I downloaded the student version.
Yesterday I removed the feedback resistor in my tube buffered amp and replaced it by a t-network, like this (gain 46.7):
The amp has now much better sound, more air... (before, it was running with 22K input-R, coupled by 4.3uF and 680K in the feedback path).
AND: it definitively solved the problem with some TIM (look here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=37488
Joe: I am sure, this way your diy circuit is more "clone proof" than with the high feedback resistor values.
Franz
Yesterday I removed the feedback resistor in my tube buffered amp and replaced it by a t-network, like this (gain 46.7):
An externally hosted image should be here but it was not working when we last tested it.
The amp has now much better sound, more air... (before, it was running with 22K input-R, coupled by 4.3uF and 680K in the feedback path).
AND: it definitively solved the problem with some TIM (look here:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=37488
Joe: I am sure, this way your diy circuit is more "clone proof" than with the high feedback resistor values.
Franz
Sorry, I cannot measure noise, but I cannot see something wrong on the scope or listen something bad.
It is a VERY good amp now. Singing.
DC at output is about 5mV, what seems to be O.K.
I get a nearby perfect square wave at 1kHz and the best square wave of all my amps at 50Hz.
With my equipment, I am not able to measure much higher frequencies than 10kHz reliable (an old Heathkit OS-2 scope, but I love it).
Franz
It is a VERY good amp now. Singing.
DC at output is about 5mV, what seems to be O.K.
I get a nearby perfect square wave at 1kHz and the best square wave of all my amps at 50Hz.
With my equipment, I am not able to measure much higher frequencies than 10kHz reliable (an old Heathkit OS-2 scope, but I love it).
Franz
Franz G said:
Good work. Not changing my input values, use 10K x 2 plus 100R = 33.3dB - I will definite want to try that.
With such low values, the 0.1uF cap in parallel with 18K (in your circuit) may not really be needed. Can you try disconnecting it and see what effect it has on the sound. Maybe no better or worse, but then again it might be better.
Can you send me an email directly to joeras@vacuumstate.com, I have a link for you that I don't want to post here. All will be revealed when I reply.
Joe R.
Joe
I would feel pleased, when this change would find a way in your DIY project!
Another point, I think, I learned: the best working point for the LM3875 is in fact a gain above 46.
Dont go below for really good sounding results. I am sure.
You get automatically some roll of at RF frequencies and the sound is not "breaked" too much by strong feedback.
Till yesterday, my OPA627 buffered gainclone (a la Carlos FM) was the winner of all my hybrids and gainclones. Today, I am not quite sure. But I cannot compare, as the Carlos Version is not here. A friend of mine, not biased, will compare them as soon as possible.
Franz
P.S.
I've sent you a mail, Joe.
I would feel pleased, when this change would find a way in your DIY project!
Another point, I think, I learned: the best working point for the LM3875 is in fact a gain above 46.
Dont go below for really good sounding results. I am sure.
You get automatically some roll of at RF frequencies and the sound is not "breaked" too much by strong feedback.
Till yesterday, my OPA627 buffered gainclone (a la Carlos FM) was the winner of all my hybrids and gainclones. Today, I am not quite sure. But I cannot compare, as the Carlos Version is not here. A friend of mine, not biased, will compare them as soon as possible.
Franz
P.S.
I've sent you a mail, Joe.
Many amps which have high open loop gain works little bit better by higher closed loop gain - read little bit more about theory Franz 😉 .
Franz G said:
Do I recall that you are using hi eff. wide range speakers? Could this have something to do with increasing the output impedence of the amp and it's effect on these drivers (see the discussion about current drive in the loudspeaker forum )?
Sheldon
Franz G said:
That is certainly on the cards.
Franz G said:
THAT is what I have been from the very beginning. Take a look at the open loop versus phase in the pdf file. The phase starts going bad - worse than 90 degrees - above 200KHz.
An externally hosted image should be here but it was not working when we last tested it.
Now note the open loop gain at that point, 30dB. If you keep the gain high you will avoid that phase lag getting into the feedback loop. I saw this very early and decided to listen for it. I believe I can hear it quite clearly. So I keep the gain above 30dB. But it does make for another problem, above 200KHz you will now have poor PSR at HF. But keeping caps close to 3875 and return the currents from the speaker load to this point, solves it nicely.
An externally hosted image should be here but it was not working when we last tested it.
Joe R.
- Status
- Not open for further replies.