OK guys, both me and Nuuk overlooked the LPF’s resistor but we’ve been through that long time ago, the accent was now on the way to calculate the effective value of the resistances used in the feedback, and to clarify possible puzzle about this. I hope the link I have provided may give the answer the people may need.
This may help calculating T-Network values versus gain:
Download CircuitMaker Student Version 6:
http://www.altium.com/circuitmaker/downloads/student.htm
Install it.
Now download this tiny zip file:
http://members.ozemail.com.au/~joeras/diy/T-Network.zip
Unpack and in CircuitMaker open this file.
You will now see a basic diagram of opamp using T-Network inverted feedback.
Double click on any resistor allows the values to be changed. Keep in mind 100k = 100k but 100 = 100 Ohm.
Now choose 'Run' from the Simulation drop-down menu or use F10.
The AC Voltmeter will then show the AC voltage out. Since it is set up with 1V RMS (+/- 1.414V) - the RMS value equals the gain, hence 46.23V means gain = 46.23
See how you go, let me know of any success.
Joe R.
Download CircuitMaker Student Version 6:
http://www.altium.com/circuitmaker/downloads/student.htm
Install it.
Now download this tiny zip file:
http://members.ozemail.com.au/~joeras/diy/T-Network.zip
Unpack and in CircuitMaker open this file.
You will now see a basic diagram of opamp using T-Network inverted feedback.
Double click on any resistor allows the values to be changed. Keep in mind 100k = 100k but 100 = 100 Ohm.
Now choose 'Run' from the Simulation drop-down menu or use F10.
The AC Voltmeter will then show the AC voltage out. Since it is set up with 1V RMS (+/- 1.414V) - the RMS value equals the gain, hence 46.23V means gain = 46.23
See how you go, let me know of any success.
Joe R.
G'day Joe,
I downloaded the zip file and unpacked it but when I try and open a file in CircuitMaker I only have the option of opening ckt or binary files. The file in the zip is an XLS! 😕
I downloaded the zip file and unpacked it but when I try and open a file in CircuitMaker I only have the option of opening ckt or binary files. The file in the zip is an XLS! 😕
Nuuk said:
I think you are trying to load Franz's XLS file?
The file in T-Network.zip is called T-Network.ckt and was created by CM Student V6.
What is the XLS full name? If it ends up T-Network.xls - then I am most definitely confused.
If so, edit to read T-Network.ckt . Also it should 4399 bytes.
Joe R.
Did anyone notice/observe/think how T-network in the feedback influences the relation between the amplitude and phase? I ran some sims with the opamp model that should match relatively well to those commonly used in the GCs (its open loop measurement shows about 3MHz bandwidth and somewhat better phase margin than LM3875’s but with generally similar shape), anyway I think it is good enough to show differences between particular configurations.
Here is its response when used in the inverted configuration and with commonly used IGC feedback of 220k/10k.
Keeping the same gain and going down with the feedback values to match the future T-network’s 10k of summed resistance between the output and the inverted input, i.e. with feedback 10k/454R, the response shows better phase, probably because the pole formed by the input capacitance and feedback resistance is shifted higher.
Now, the same gain and T-network in the feedback (5k/118.8R/5k). Obviously, response disappears earlier, leaving the phase at new unity gain point still better.
Pedja
Here is its response when used in the inverted configuration and with commonly used IGC feedback of 220k/10k.
An externally hosted image should be here but it was not working when we last tested it.
Keeping the same gain and going down with the feedback values to match the future T-network’s 10k of summed resistance between the output and the inverted input, i.e. with feedback 10k/454R, the response shows better phase, probably because the pole formed by the input capacitance and feedback resistance is shifted higher.
An externally hosted image should be here but it was not working when we last tested it.
Now, the same gain and T-network in the feedback (5k/118.8R/5k). Obviously, response disappears earlier, leaving the phase at new unity gain point still better.
An externally hosted image should be here but it was not working when we last tested it.
Pedja
Pedja said:
Good work Pedja.
So the phase is straighter because, with the lower Z we're more able to drive the input capacitance. That sounds rather rational.
So I got the idea, this is an opamp capable of driving Lo Z speaker loads, hence in a T network can be made really low as it will have no difficulty driving that. So how about making the T 2 x 1K (make these 1 Watt) and One Ohm. With my 4k7 + 18K input Z (22K7 total), this gives a gain of 45.
This not gets maximum drive but should also give low DC offset.
Joe R.
carlosfm said:
No, but try to give a different name.... you will just end up with a bloody nose!
carlosfm said:
For quite some time, yes. As long as people call the things with certain names, it is smart to do the same, in the things of language you can’t go against the stream. (It is not that I am proposing this as a universal rule, it is about the language.) Of course, during the time the term gainclone may fade.
Joe was faster sending the message... 😉Joe Rasmussen said:
No, but try to give a different name.... you will just end up with a bloody nose!
Problem solved Joe. For some odd reason, I ahd downloaded Franz's file to my desktop and then downloaded yours. The Franz zip appeared but your didn't although it was shown as being in my Desktop Folder. 😕
Anyway, when I did a search for your zip file and clicked on the result, it unzipped and I can now view it! 🙂
Re the name thing. I bet most of us don't really like the name Gainclone but as a previous thread revealed, no matter how many better alternatives are put forward, we always get stuck with the Gainclone tag.
Like Pedja says, even if some of us start referring to them by a different name, all that will do is cause confusion.
To rememdy this situation I suggest a World Chip Amp Conference to be held somewhere exotic (Laverton?) where we will all meet and agree on a new name over a few beers! 😉
Anyway, when I did a search for your zip file and clicked on the result, it unzipped and I can now view it! 🙂
Re the name thing. I bet most of us don't really like the name Gainclone but as a previous thread revealed, no matter how many better alternatives are put forward, we always get stuck with the Gainclone tag.
Like Pedja says, even if some of us start referring to them by a different name, all that will do is cause confusion.
To rememdy this situation I suggest a World Chip Amp Conference to be held somewhere exotic (Laverton?) where we will all meet and agree on a new name over a few beers! 😉
Thanks, Joe.Joe Rasmussen said:
What seems interesting is, else than this benefit of lower resistance, the opamp acts effectively like lower bandwidth unit, like the unit having another nice compensation pole with better driving capabilities.
Yes, good idea, there is a nice possibility of setting resistors values there practically independently of the input resistor. I think the second 1k and 1R should be one watt, the power rating of the first 1k is not critical.
Pedja
Very interesting, what happens with a lower feedback Z!
And the sound is great 😀
Tonight, I will convert my regulated BIGC from 10K/220K to 10k/1K/2.3R/1K, gain -43.6
I think, just one 10K, this at the output, must be 1 Watt.
Franz
And the sound is great 😀
Tonight, I will convert my regulated BIGC from 10K/220K to 10k/1K/2.3R/1K, gain -43.6
I think, just one 10K, this at the output, must be 1 Watt.
Franz
Franz, can you measure distortion ? Can you show us photo of 10 kHz into load, by 1 and 10 Watt ?
Yes, I must correct myself, considering the power rating, 1R won't be critical, it is small voltage across it, so only the second series resistor in the T-network will be.
There may be a problem here. Some offset voltage which is the sum of the input offset voltage and bias current multiplied by the first series resistor will appear in the middle point of the T network feedback and this voltage will be amplified by the ratio between the second series and shunt resistor of the T-network.
I assumed non-inverting input shorted to the ground and tried a bit of calculating with the LM3875’s 1mV input offset voltage and 0.1uA bias current (something like this is more typical for it than 0.2uA claimed in the datasheet), but have not finished with any lower offset then I have with conventional IGC, actually it is easy to finish with notably higher values.
I think the usage of the resistor from the non-inverting input to ground (no cap in parallel) is in this case the way to go, both low DC offset and the same or similar impedances seen by both inputs can be achieved at the same time. However, I think the things won’t go much far from the schematic you have already posted.
T network like 1k/1R/1k possibly can be using a trimmer between the non-inverting input and ground.
Pedja
Joe Rasmussen said:
There may be a problem here. Some offset voltage which is the sum of the input offset voltage and bias current multiplied by the first series resistor will appear in the middle point of the T network feedback and this voltage will be amplified by the ratio between the second series and shunt resistor of the T-network.
I assumed non-inverting input shorted to the ground and tried a bit of calculating with the LM3875’s 1mV input offset voltage and 0.1uA bias current (something like this is more typical for it than 0.2uA claimed in the datasheet), but have not finished with any lower offset then I have with conventional IGC, actually it is easy to finish with notably higher values.
I think the usage of the resistor from the non-inverting input to ground (no cap in parallel) is in this case the way to go, both low DC offset and the same or similar impedances seen by both inputs can be achieved at the same time. However, I think the things won’t go much far from the schematic you have already posted.
T network like 1k/1R/1k possibly can be using a trimmer between the non-inverting input and ground.
Pedja
Pedja said:
This may work fine with these chips, but some lower bandwidth chips like the OPA549 may not work with this topology, at these gains.
I prefer the LM chips, anyway.😀
What about T-Amp, isn't it easyer to call it this way than a GC?
It's just that there are so many different implementations of the "GC" that to say what you mean you need to write 10 words.😀
T-Amp seams😎 .
Joe Rasmussen said:
Joe,
The ideal topology will present exactly the same
R and C impedances to each (+ and -) input.
This will minimise CM distortion and also offset.
It will also sound better, we have tried this on
opamps and it works.
When looking at each RC network treat the opamp OP as
ground.
Cheers,
Terry.
Sheez,
This is worse than computers a few years ago, when the new models were obsolete by the time they hit the shelfs. My solder joints aren't even cool and the last batch of parts I ordered haven't come in, and you guys come up with new things to try. What's a compusive tinkerer to do? It may be that the only way for me to get any listening in is to avoid the forum for a few days at a time.
Sheldon
By the way, with a low value T network (1k,2.3R,1k) what happens to the requirement for the Zobel? Seems like it would be less critical.
This is worse than computers a few years ago, when the new models were obsolete by the time they hit the shelfs. My solder joints aren't even cool and the last batch of parts I ordered haven't come in, and you guys come up with new things to try. What's a compusive tinkerer to do? It may be that the only way for me to get any listening in is to avoid the forum for a few days at a time.
Sheldon
By the way, with a low value T network (1k,2.3R,1k) what happens to the requirement for the Zobel? Seems like it would be less critical.
Sheldon - two systems, two rooms.
Have one system for listening to and another for tinkering with. That way if you get tired of making a mod, you just go to the other room and enjoy listening to the music. 😉
- Status
- Not open for further replies.