oops. 😱
Sorry to all, carlosfm's suggested values were 15k, 18k-1K-18k, giving a gain of 24. I was going to test 22k, 22k-1k-22k (also a gain = 24).
I have been using my simple spreadsheet that uses the formula from "Op amps for Everyone" by Ron Mancini. 😉
Sorry to all, carlosfm's suggested values were 15k, 18k-1K-18k, giving a gain of 24. I was going to test 22k, 22k-1k-22k (also a gain = 24).
I have been using my simple spreadsheet that uses the formula from "Op amps for Everyone" by Ron Mancini. 😉
i couldn't find my old post either. it was a long time ago now. i do actually have the amp sitting next to me though. its a jumble of wires, a 1.2" x 2.2" pcb, and a heatsink...
red brown black black yellow yellow? hmm, 1% codes are odd...
red black black red brown?
i'm gonna have to take the amp off the heatsink...
from the output of amp to the T junction i get 32.6k (maybe it was 33k?). from T to ground was 329ohm (probably 330ohm). T to inverting was 19.6k (probably 20k). input was 99.8k (likely 100k) 1uf cap on input.
can't find the impedance from non-inveritng to ground, but IIRC it was near 20k. i think this gave a gain between 20 and 30.
red brown black black yellow yellow? hmm, 1% codes are odd...
red black black red brown?
i'm gonna have to take the amp off the heatsink...
from the output of amp to the T junction i get 32.6k (maybe it was 33k?). from T to ground was 329ohm (probably 330ohm). T to inverting was 19.6k (probably 20k). input was 99.8k (likely 100k) 1uf cap on input.
can't find the impedance from non-inveritng to ground, but IIRC it was near 20k. i think this gave a gain between 20 and 30.
theChris said:from the output of amp to the T junction i get 32.6k (maybe it was 33k?). from T to ground was 329ohm (probably 330ohm). T to inverting was 19.6k (probably 20k). input was 99.8k (likely 100k) 1uf cap on input.
Thanks theChris, using your values 100k, 20k-330-33k gives a gain of 20.5. Can you confirm you don't have any hiss?
this was done for a subwoofer with PC soundcard source, so i didn't pay much attention. i don't recall any notciable hiss though, it seemed perfectly quiet.
theChris
About the hissing, carlosfm has suggested that these chips start to get noisy at around 24dB of gain and there has been evidence to support this axiom but there are other members who have built the ITGC with gain over 30dB and have had quiet amps. This seems to indicate that there is either a pretty sizable performance variance in these chips or the buildout is a strong factor in determining how much gain you can get away with before making noise.
About the hissing, carlosfm has suggested that these chips start to get noisy at around 24dB of gain and there has been evidence to support this axiom but there are other members who have built the ITGC with gain over 30dB and have had quiet amps. This seems to indicate that there is either a pretty sizable performance variance in these chips or the buildout is a strong factor in determining how much gain you can get away with before making noise.
This seems to indicate that there is either a pretty sizable performance variance in these chips or the buildout is a strong factor in determining how much gain you can get away with before making noise.
Or that speakers/crossovers are an issue too!
Nuuk
Please explain. I had assumed that the buildout would take into account the speaker system. Let's examine the amp circuit interaction with the speaker system. Are we talking about:
(a) speaker impedence
(b) amp circuit interaction with crossover
(c) speaker sensitivity
Please explain. I had assumed that the buildout would take into account the speaker system. Let's examine the amp circuit interaction with the speaker system. Are we talking about:
(a) speaker impedence
(b) amp circuit interaction with crossover
(c) speaker sensitivity
Well for instance, the sensitivity of the driver will play a part, also whether the driver is connected directly to the amp or through a crossover.
In other words, the more revealing your speakers are, the more you will notice the hiss.
In other words, the more revealing your speakers are, the more you will notice the hiss.
Okay, I agree that sensitive speakers have a lower noise floor and will pick up hiss that less sensitive speakers ignore. A speaker with 100dB sensitivity could have hiss that a 96dB sensitive speaker will ignore but that doesn't explain how hiss from a speaker impedence curve or crossover can be attributed to the amp gain. Please exclude the case of active crossover speakers from your reply.
Please exclude the case of active crossover speakers from your reply.
I wasn't giving a definitive answer, just adding in another possible factor! 😉
I'm a bit mystified about the TN hiss as well. The first time I tried it - zero hiss. The last time, same values, quite a bit of hiss that was reduced by lowering the gain. 10K-10K with 220R to ground instead of 100R.
Nuuk
...and this observation was with the same chip?! If it was then that is puzzling but if you changed the chip it affirms the point I brought up earlier about chip variance.
...and this observation was with the same chip?! If it was then that is puzzling but if you changed the chip it affirms the point I brought up earlier about chip variance.
Just some info based on my TIGC that I build (LM3875).
I had the classic NIGC and converted it to TIGC. The values I used: 100k (+in to GND), 100k(-in to pot), 100k-1k-22k (FB divider). This gives you gain of 22 and Zin o 100k. The output offset were -13 and 6 mV.
There was hiss and even 60Hz hum. Sound: the base was pronounced and a touch too lose but not unpleasant. Putting cap across the 100k (+in to gnd) reduced the hiss and the hum. It looked like the +in pin likes lower resistance to GND.
Changed the values as follows:
10k (+in to GND), 47k(-in to pot), 10k-100-10k (FB divider). Gain of 21. Hiss was further reduced and the hum reduced dramatically. It seams that the base sounds more controlled. The output offset, though went to +20 and -60 mV. Change the resistors (+in to GND) with 3.3k instead of 10k to compensate for the offset and now it's at -20 and + 20 mV. In this configuration the LM has a gain of 100 and the whole amp has gain of 21, so the hiss got reduced by lowering the res +in to GND and increased again by the much higher gain of the IC.
/Greg
I had the classic NIGC and converted it to TIGC. The values I used: 100k (+in to GND), 100k(-in to pot), 100k-1k-22k (FB divider). This gives you gain of 22 and Zin o 100k. The output offset were -13 and 6 mV.
There was hiss and even 60Hz hum. Sound: the base was pronounced and a touch too lose but not unpleasant. Putting cap across the 100k (+in to gnd) reduced the hiss and the hum. It looked like the +in pin likes lower resistance to GND.
Changed the values as follows:
10k (+in to GND), 47k(-in to pot), 10k-100-10k (FB divider). Gain of 21. Hiss was further reduced and the hum reduced dramatically. It seams that the base sounds more controlled. The output offset, though went to +20 and -60 mV. Change the resistors (+in to GND) with 3.3k instead of 10k to compensate for the offset and now it's at -20 and + 20 mV. In this configuration the LM has a gain of 100 and the whole amp has gain of 21, so the hiss got reduced by lowering the res +in to GND and increased again by the much higher gain of the IC.
/Greg
...and this observation was with the same chip?! If it was then that is puzzling but if you changed the chip it affirms the point I brought up earlier about chip variance.
Yes different chips (both LM3875's). So this could be a signifcant factor with the T-network as well as affecting the DC offset.
greg, the +in to ground will be usually close to:
resistors: inv to T + T to gnd
becasuse this will be the shorted path and all other resistors will be "large"
it would be:
(Rti + Rtg || Rto ) || Ri
and since Rto >> Rtg and Ri >> Rti + Rtg, the equation simplifies to just Rti + Rtg.
"Please exclude the case of active crossover speakers from your reply."
--i would think the exact opposite. a passive crossover could mask the noise if the amp was used only on the woofer as in my case.
btw:
what was the gain formula again?
resistors: inv to T + T to gnd
becasuse this will be the shorted path and all other resistors will be "large"
it would be:
(Rti + Rtg || Rto ) || Ri
and since Rto >> Rtg and Ri >> Rti + Rtg, the equation simplifies to just Rti + Rtg.
"Please exclude the case of active crossover speakers from your reply."
--i would think the exact opposite. a passive crossover could mask the noise if the amp was used only on the woofer as in my case.
btw:
what was the gain formula again?
yldouright said:theChris
About the hissing, carlosfm has suggested that these chips start to get noisy at around 24dB of gain and there has been evidence to support this axiom but there are other members who have built the ITGC with gain over 30dB and have had quiet amps. This seems to indicate that there is either a pretty sizable performance variance in these chips or the buildout is a strong factor in determining how much gain you can get away with before making noise.
Guessing, guessing, guessing...🙄
I didn't say that.
What I said I will transcribe here, it was a suggestion:
"...gain of 24, fine for a power amp.
For an integrated amp I would use a gainstage (2x~2.5x)."
Does this mean that I said that above this gain you will have hiss?
No.
You will notice hiss above ~40x gain, and most t-networkers are using gains at around 45.
My suggestion, and you know that, was to use the chip at a "normal" gain, because it was made for power amps, and use a gainstage before it.
It also sounds better this way, and it is dead silent, of course.
A signal op-amp is much more silent than a power op-amp, which was not made to be used alone at high gains.
My pre has a gain of 3x.
My power amp has a gain of 22x.
This is 66x total, quite "standard".
Try to give this gain to an LM power op-amp alone and you'll know what I mean.
Hiss? What hiss? I use gain of 46 (LM3875).
My suggestions, just a brainstorming:
- use small bypass caps nearby the chips legs (0,1uF)
- try bypassing the rectifier diodes by small caps (0.01uF to 0.1uF, work it out).
- use a rf filter before the tranny or at least a 100nF cap in parallel to the primary windings.
- CORRECT grounding, from A to Z
- check for every topology problem
- last but not least: check for every possible rf oscillation in the circuit!
O.K., there is a variation in chips, and I can hear some hiss nearby (10 cm) my 100dB/W speakers.
But definitely not at the listening position!
Franz
My suggestions, just a brainstorming:
- use small bypass caps nearby the chips legs (0,1uF)
- try bypassing the rectifier diodes by small caps (0.01uF to 0.1uF, work it out).
- use a rf filter before the tranny or at least a 100nF cap in parallel to the primary windings.
- CORRECT grounding, from A to Z
- check for every topology problem
- last but not least: check for every possible rf oscillation in the circuit!
O.K., there is a variation in chips, and I can hear some hiss nearby (10 cm) my 100dB/W speakers.
But definitely not at the listening position!
Franz
Franz G said:Hiss? What hiss? I use gain of 46 (LM3875).
My suggestions, just a brainstorming:
- use small bypass caps nearby the chips legs (0,1uF)
- try bypassing the rectifier diodes by small caps (0.01uF to 0.1uF, work it out).
- use a rf filter before the tranny or at least a 100nF cap in parallel to the primary windings.
- CORRECT grounding, from A to Z
- check for every topology problem
- last but not least: check for every possible rf oscillation in the circuit!
O.K., there is a variation in chips, and I can hear some hiss nearby (10 cm) my 100dB/W speakers.
But definitely not at the listening position!
Franz
When I referred to hiss and hum (in my previous post) you could hear only at 5-10 cm distance from the speakers. At listening position you cant hear anything.
Yes, the hiss was only audible close to the speakers in my case. But I think that we should aim for total silence"
Franz G said:But definitely not at the listening position!
Of corse not, Franz!
But I have my bech system right in front of me and I couldn't bear to listen to that hiss, as I listen to this at very low volumes.
Not hum, hiss.
I had a gain of 44x (LM1875, t-network), with an input buffer.
I changed the (noisier) LM1875 for 22x gain and the input stage (OPA2132) was changed for a gain of ~2x.
Dark silence from minimum to max volume.😎
It also sounds better.😉
Nuuk said:Yes, the hiss was only audible close to the speakers in my case. But I think that we should aim for total silence"
I do.
Welcome to the club.

- Status
- Not open for further replies.
- Home
- Amplifiers
- Chip Amps
- T-network: the better feedback solution?