T-network: the better feedback solution?

Status
Not open for further replies.
For gengcard- I tried this with my (unbuffered) inverted OPA549 amp.I used JoeR's parts values-10k+100r+10k and 10k on the + input. This does work, but I ended up with a huge amount of hiss thru the speakers.Don't know why it turned out this way-maybe the input buffer is nesessary for this mod? Went back to the 220k feedback with 220k on the + input. Sounds just as good and about 98% less noise!
 
mrskinny said:
For gengcard- I tried this with my (unbuffered) inverted OPA549 amp.I used JoeR's parts values-10k+100r+10k and 10k on the + input. This does work, but I ended up with a huge amount of hiss thru the speakers.Don't know why it turned out this way-maybe the input buffer is nesessary for this mod? Went back to the 220k feedback with 220k on the + input. Sounds just as good and about 98% less noise!


Hi!
I think this ist way to much gain for the opa549.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=40189&highlight=

Regards

Tor Martin
 
GregGC said:



Thanks Joe.
Because the I-offset is the problem, want would be the disadvantage then to still do what I proposed (DC gain of 5-10 and add AC gain with 47k FB res.) and fine-tune the resistor from +in to GND (not injecting voltage) to get the Output DC offset to zero. It sounds like the simplest way to solve the Offset issue. It'll create only a minor Z imbalance which shouldn't be a problem.

/Greg

That's exactly what we (at least four that I know off) were trying to avoid. Oh yes,I tried it.

The aim here was to keep the R & C the same and come up with a set of R values that will cope with a wide variety of offsets and yet (hopefully) have around or below 50mV DC offset.

The R value from (+) to gnd must match for best sound. If you use 47K in conjunction DC gain as low as you propose will not create a minor Z imbalance you suggest. It will be major. Indeed with 10K/100R/10K & 10K on (+) and 100 DC gain, to trim the DC offset may require the 10K to (+) to be trimmed as much as +/- 20%. If you use 47K base value and low DC gain, this will be even worse.

Also, please consider this: It's fine for us to come up with various esoteric one-off solutions, but there are those reading this who wouldn't be able to calculate AC gain of T-networks, let alone cope with major potential DC shifts it creates.

So: Rather than speculate I decided to spend a number my hours on the weekend, to come up with a VW solution... you know, the car for anybody who needs a car solution. This is more helpful than one-offs and maybe just as good. So we are not talking VW performance - just VW accessibility.

Sound wise it works and now anybody can try it.

Joe R.
 
Shoog said:
Hi there,
I have been following the posts about DC offset with interest. I noticed a higher DC component after the implementation of the T-network, but it dropped with a resistor to the non-inverting input. These measurements where made at idle. Am I to understand that the DC is created when the amp is been driven, and not when at idle. If so has anyone actually observed it as a problem ?? How much DC offset are we talking about ???

Help a poor soul to understand.

Shoog

Just measure the DC statically, no sound.

Can you post the values and the DC offset in mV. for both channels. This will be helpful feedback especially if you used the 10K based T-network. I have measured six 3875s with those values and so far below 50mV. I hope yours follow the same pattern.

BTW, The DC is always there, the AC is not. When AC settles (the sound dies down) the DC offset remains. The DC becomes the 0V reference for the AC. But the remaining DC will always push current through the Voice Coil of the driver, pulling the driver of centre (either slightly in or out - just like a battery does when applied to a speaker - try it), it also means the amp is wasting power that becomes expended as heat in both itself and the voice coil. So the DC needs to be kept low. Hope that helps.

Joe R.
 
Hi mrskinny,


For gengcard- I tried this with my (unbuffered) inverted OPA549 amp.I used JoeR's parts values-10k+100r+10k and 10k on the + input. This does work, but I ended up with a huge amount of hiss thru the speakers.Don't know why it turned out this way-maybe the input buffer is nesessary for this mod? Went back to the 220k feedback with 220k on the + input. Sounds just as good and about 98% less noise!

O.K. give up OPA549 for T-network a moment. 🙁 Let me start with LM3875 firstly. 😀
 
mrskinny said:
For gengcard- I tried this with my (unbuffered) inverted OPA549 amp.I used JoeR's parts values-10k+100r+10k and 10k on the + input. This does work, but I ended up with a huge amount of hiss thru the speakers.Don't know why it turned out this way-maybe the input buffer is nesessary for this mod? Went back to the 220k feedback with 220k on the + input. Sounds just as good and about 98% less noise!

Mr skinny

WRT OPA549 noise:

Read the data sheet WRT IP voltage noise (70nV/rt Hz) and
current noise (1pA/rt Hz). Download one of the many application
notes from AD and work out the OP noise based on the IP
impedances and gain structure.

The current noise will interact with IP Z and be multiplied
by the gain of T network (100). Voltage noise will be RMS
summed to this also. At the end of it things get pretty
noisy just looking at some quick calcs.

It may also be oscillating, did you check this?


Cheers

Terry
 
Joe Rasmussen said:



The aim here was to keep the R & C the same and come up with a set of R values that will cope with a wide variety of offsets and yet (hopefully) have around or below 50mV DC offset.

The R value from (+) to gnd must match for best sound. If you use 47K in conjunction DC gain as low as you propose will not create a minor Z imbalance you suggest. It will be major. Indeed with 10K/100R/10K & 10K on (+) and 100 DC gain, to trim the DC offset may require the 10K to (+) to be trimmed as much as +/- 20%. If you use 47K base value and low DC gain, this will be even worse.

Joe R.

Joe,

Stop wasting time postulating on different gain
structures. What we really need to ascertain is how much
effect balancing the impedances to each IP makes *sonically*.
Usually it does improve sonics WRT opamps, especially
bipolar IP, however we don't know in this case.

My fully balanced / parrallel 3886 module is about 1/2 done.
It has exactly balanced everything, with a separate discrete
0 FB front end (with gain adj R).

The design is direct coupled IP to OP and hopefully offset
will not be a problem. Should be extremely linear but real proof
is in the listening so your JLTI is a good reference.

I'll give you a holler when it's done (hopefully by next week).

Unfortunately I will not get AP test set for a while so maybe
we can try your cleo.

Cheers,

Terry
 
Joe Rasmussen said:


That's exactly what we (at least four that I know off) were trying to avoid. Oh yes,I tried it.

The aim here was to keep the R & C the same and come up with a set of R values that will cope with a wide variety of offsets and yet (hopefully) have around or below 50mV DC offset.

The R value from (+) to gnd must match for best sound. If you use 47K in conjunction DC gain as low as you propose will not create a minor Z imbalance you suggest. It will be major. Indeed with 10K/100R/10K & 10K on (+) and 100 DC gain, to trim the DC offset may require the 10K to (+) to be trimmed as much as +/- 20%. If you use 47K base value and low DC gain, this will be even worse.

Also, please consider this: It's fine for us to come up with various esoteric one-off solutions, but there are those reading this who wouldn't be able to calculate AC gain of T-networks, let alone cope with major potential DC shifts it creates.

So: Rather than speculate I decided to spend a number my hours on the weekend, to come up with a VW solution... you know, the car for anybody who needs a car solution. This is more helpful than one-offs and maybe just as good. So we are not talking VW performance - just VW accessibility.

Sound wise it works and now anybody can try it.

Joe R.


Thanks Joe!

Nice of you to give explanation. Sorry to waste your time.

/Greg
 
Hi Terry

Terry_Demol said:

Stop wasting time postulating on different gain structures. What we really need to ascertain is how much effect balancing the impedances to each IP makes *sonically*. Usually it does improve sonics WRT opamps, especially bipolar IP, however we don't know in this case.

WRT LM3875, I think we do. The answer is positive.

My fully balanced / parrallel 3886 module is about 1/2 done. It has exactly balanced everything, with a separate discrete
0 FB front end (with gain adj R).

Getting interesting, especially what sort of Watts out. My Dummy Load is rated @ 300W - see if we can put a bit of heat into it? Re gain, take it for granted, needs to be about x45. Best for sound and also for my -11dB output on the Sony 555. Oh, I just remembered, you haven't heard this latest guise of my 555. Even Steve and Dragon was impressed, very!

The design is direct coupled IP to OP and hopefully offset will not be a problem. Should be extremely linear but real proof is in the listening so your JLTI is a good reference.

With cryo-ed BBs and T-network - it's some reference to match. You never heard the JLTi sing like this. BTW, spoke to Steve earlier today, in 2-3 weeks time he reckons he will be able to cryo tubes - as he's getting the equipment together. As Big Kev would say: "I'm excited!"

I'll give you a holler when it's done (hopefully by next week). Unfortunately I will not get AP test set for a while so maybe
we can try your cleo.

C*L*I*O* - not a bad backup to AP. The relative noise floor as measured in Clio is lower with T-network. I posted the results earlier on in this thread. Did you see them? Very nice.

Joe R.
 
Joe Rasmussen said:


Hi Greg

I don't think of it as a waste of time, so don't worry - it gave an opportunity to present matters clearly. Those of us promoting T-networks would like as many as possible to try it, and now they can without trepidation.

Joe R.

OK, I'll retract my statement, it was probably a bit
in haste 🙂

It is certainly a credit to you to spend so much time
with these posts helping people out as I know how
busy you are.

Cheers,

Terry
 
Joe Rasmussen said:
Hi Terry

WRT LM3875, I think we do. The answer is positive.

OK, so you have done direct comparison and balancing
impedances is better?

Getting interesting, especially what sort of Watts out. My Dummy Load is rated @ 300W - see if we can put a bit of heat into it? Re gain, take it for granted, needs to be about x45. Best for sound and also for my -11dB output on the Sony 555. Oh, I just remembered, you haven't heard this latest guise of my 555. Even Steve and Dragon was impressed, very!

I can arrange the IP gain very easily. It is independant of
LM3886 gain which runs fixed. The IP merely feeds current
into the 3886. However I'm not convinced that if properly
compensated and balanced that x45 is optimum.
I will start low (x15) and work up. There is much to do
as yet and much is unknown at this stage. We will see.

With cryo-ed BBs and T-network - it's some reference to match. You never heard the JLTi sing like this. BTW, spoke to Steve earlier today, in 2-3 weeks time he reckons he will be able to cryo tubes - as he's getting the equipment together. As Big Kev would say: "I'm excited!"

I don't expect to match that setup as I believe that a simple
3875 set up SE is probably superior to a higher power bridge
parrallel circuit, but the application is for studio monitoring
where 100W minimum into 8 ohms required. So that is fixed.

C*L*I*O* - not a bad backup to AP. The relative noise floor as measured in Clio is lower with T-network. I posted the results earlier on in this thread. Did you see them? Very nice.

Joe R.

I'll check them out. The advantage of power amps is you just
crank up the power and the non linearities will be easily
seen, it's not like we are trying to measure a 627 or such 🙂

Cheers,

Terry
 
NIGC

Anyone out there tried this t-network on the LM3875-based non-inverted gainclone?

For those using the IVBGC, was the bass performance of the LM3875 amp the same before and after the incorporation of the VB?

My findings and the findings of a couple of magazine reviewers is that the use of a pre-amp with the LM3875 amp results in a slight roll-off in the bass performance of the LM3875 relative to the use of no preamp.
 
My findings and the findings of a couple of magazine reviewers is that the use of a pre-amp with the LM3875 amp results in a slight roll-off in the bass performance of the LM3875 relative to the use of no preamp.

Passive pre and 25 metre long interconnects Carlos! 😀

I don't observe this at all with my new (active) GC pre amp/monoblock combination! 😉
 
Status
Not open for further replies.