Hi,
I am in favour of Transformers for Balanced Interfaces in general, inputs, outputs etc.
Well, first THD at low frequencies is not "MUCH WORSE" if you competently select the transformer, secondly, THD as such is one of the most meaningless Amplifier Specifications, together with Damping Factor. In the real world neither number has ANY bearing to the sound quality of a given amplifier.
However, the natural and passive bandwidth limiting behaviour of line input transformers can reliably protect the Amplifier input from ultrasonic noise that will cause signal dependent noisefloor modulation (a highly relevant factor to amplifier sonics not captured by ANY measurements) and provides a reliable and simle solution to a universal (balanced, unbalanced) input.
Of course, all this CAN be done electronically. To do it right requires several high performance op-amps, many passive components and still does not allow galvanic insulation, which is the only reliable way to kill noise loops.
As long as you are not bothered about interconnecting multiple pieces of equipment in the real world, with plenty of RF soup around from anything starting with VLFW from comms to nuke subs to your cellphone, with plenty of mainsborne noise and the like, as long as you connect the Amp in a lab to a AP One, instead of working it out under realistic conditions you will not understand the reason why you MUST use transformers if fidelity of the signal is desired.
Sayonara
I am in favour of Transformers for Balanced Interfaces in general, inputs, outputs etc.
Well, first THD at low frequencies is not "MUCH WORSE" if you competently select the transformer, secondly, THD as such is one of the most meaningless Amplifier Specifications, together with Damping Factor. In the real world neither number has ANY bearing to the sound quality of a given amplifier.
However, the natural and passive bandwidth limiting behaviour of line input transformers can reliably protect the Amplifier input from ultrasonic noise that will cause signal dependent noisefloor modulation (a highly relevant factor to amplifier sonics not captured by ANY measurements) and provides a reliable and simle solution to a universal (balanced, unbalanced) input.
Of course, all this CAN be done electronically. To do it right requires several high performance op-amps, many passive components and still does not allow galvanic insulation, which is the only reliable way to kill noise loops.
As long as you are not bothered about interconnecting multiple pieces of equipment in the real world, with plenty of RF soup around from anything starting with VLFW from comms to nuke subs to your cellphone, with plenty of mainsborne noise and the like, as long as you connect the Amp in a lab to a AP One, instead of working it out under realistic conditions you will not understand the reason why you MUST use transformers if fidelity of the signal is desired.
Sayonara
I think he just did that?Bricolo said:
Klaus
The extra resistor is a current bias-compensation resistor. It reduces the current bias by eliminating non-zero current at the inputs. Look also http://www.play-hookey.com/analog/experiments/basic_op_amp_inverter.html
About the cap I am not sure –should reduce power supply noise at non-inverting input?
Got my inverted clone up and running last night. I followed Peter Daniel’s minimal circuit –so only two resistors and input capacitor. Building this amp is well worth effort IMHO.
Sound is nice and polite not tiring, easy to listen, slightly romantic (as somebody else already mentioned). Details are all there, but not with extreme nicety to sidetrack your attention from overall music flow. Bass is a bit shy and low midrange could benefit by having more authority and darkness around. I think that it is because of small filter capacitors(1000uF) - will try 2200uF instead. Chips are cold as dead cats.
So far all is well.
The problems I encountered is PS hum (buzz) in output and that amp is extremely sensitive to mains disturbances (like from broken door bell or refrigerator activating)and RFI.
Circuit and PS grounds are all running into one small ground point (solder joint formed by all ground return wires). Inputs and outputs are isolated from chassis.
Could be that this too tight ground point causing the hum? Could bypass cap from ground to non-inverting input(as in Thorsten’s original schematic) solve the hum case. Could zobel at the output provide immunity to the outside radio noises?
Cheers,
Argo
I went out to my parents this weekend and hooked my inverted prototype(deadbug?) to my father's ESS AMT monitors. Not particularly a difficult load but tougher than anything I have and has better detail. The amp is almost dead quiet and sounded very nice. My father was very impressed once we compared it to all of his equipment (Dynacos, Nak TA4, and a few other various combinations). While it does have excellent range and staging, it does lack the real low end slam that larger discrete amps have, but the detail was there as well as the midrange warmth. We ran the amp hard for 2 hours and the 3"x4" x1/4" pieces of Al the ICs were screwed to never got warmer than 115degsF. I kept a .16ohm output resistor in to keep the amp stable with different loads, as well as a 2.0uF cap and 10k resistor on the input leg.
One question came up that maybe Kuei can answer... Once the filter caps are charged and I turn the amp off then back on I get a loud thump that I don't get went the amp has sat off for a while. I would have thought it would be the other way around because of the inrush from the filter caps charging up. I am not using the RC circuit to ground on the positive input, just shorted. Is this from not using any impedence on that leg and the charge on the caps just sitting on the input legs?
One question came up that maybe Kuei can answer... Once the filter caps are charged and I turn the amp off then back on I get a loud thump that I don't get went the amp has sat off for a while. I would have thought it would be the other way around because of the inrush from the filter caps charging up. I am not using the RC circuit to ground on the positive input, just shorted. Is this from not using any impedence on that leg and the charge on the caps just sitting on the input legs?
Thanks, Argo, for the explanation and the link - I just came back from work (after midnight and after 4 hours talking to sleepy students and 2 x 400km by train), I will read it later today.
I am a bit puzzled because there is then such a variety of values possible: works with a short and a 220k resistor ?!
My new inverted chip-amp sounded very, very good from the start, and I must admit that I really was a bit sceptic before and I did not expect much compared to the "normal" non-inverted circuit. I am still testing, and I am a bit unsure to write down clearly defined expressions about the sound yet, because it sounds so much different, in a way, to more classical high-quality hifi amplifiers, like my Naims.
Polite? Not in the way I would call something polite, more neutral, a bit like a perfect butler with no own opinion.
Not tiring ? Yes, definitely. I heard so many of this very clean sounding amps, which tired me to death (like Krells, ML, etc. - sorry folks).
Bass shy ? I thought that at first glance but after putting on records with lots of real bass I could not confirm it. Bass is maybe a bit more right ? My inverted chip-amp can play really loud (24V secondaries), and, although running hot then, shows no signs of giving up, and even the bass is very stable. This fact amazes me most, because the PSU is a slap in the face of a normal textbook one.
I already tried a 2 x 2200yF version (non-inverted), which sounded quite similar, but much more in the usual hifi sense. It is hard to describe: I just did not like that much than the 1000yf version (there were other slight differences too, I must admit, like metal films, FB ratio 20/1 and a NFB CD blocking cap).
The musical flow is there and well preserved but the presentation of details is quite different. But this is still just a test and not a defintive setup. I will try different input caps, an impedance smoothing resistor (.22 Ohm, like suggested), and other mods. But this amp is much "faster" than many others, the sound starts and stops in an breathtaking both secure and lively way, It is truly amazing.
I can confirm your observations about the sensitivness of the inverted version to mains noise. Although my amp is as quiet as its brothers (or sisters) and completetly noise and hum-free, it picks up mains noise quite heavily like the switching noise of the *******ing fridge from the kitchen nearby.
I read in another thread here on this forum about a very possible reason: In the inverted version the input (output of the source and cable) is a part of the feedback loop and therefore very sensitiv to picking up all kind of noise. In every mixing desk the most sensitive point is the input of the mixing amp connected to the mixing rails (usually a virtual ground inverted amplifier). Here normally every possible care is taken to sreen everything properly. Very expensive mixing desks therfore even have symmetric (balanced) mixing rails to get rid of the least bit of RFI noise picked up at this parts of the circuit.
Thor/Kwei was suggesting here (or elsewhere) to ground the resistor (or the short maybe) to the input ground first, not to the more noisy star ground between the psu caps. I did not do that, I soldered everything to the starground - I mean the whole circuit is about a square centimeter or less, so I did not see much benefit form that. But maybe ?
A solution, already suggested in this thread would be an input buffer outside of the feedback loop or a well designed preamp.
(That is the path the Final amps are following IMO. Did somebody try to reengineer these ones already ? The main thing seems to be, besides the odd dry battery power psu, a low gain preamp with low Z output, and an inverted LM1875 circuit with a two-part feedback resistor, one paralleled with a pot.)
Klaus
I am a bit puzzled because there is then such a variety of values possible: works with a short and a 220k resistor ?!
My new inverted chip-amp sounded very, very good from the start, and I must admit that I really was a bit sceptic before and I did not expect much compared to the "normal" non-inverted circuit. I am still testing, and I am a bit unsure to write down clearly defined expressions about the sound yet, because it sounds so much different, in a way, to more classical high-quality hifi amplifiers, like my Naims.
Polite? Not in the way I would call something polite, more neutral, a bit like a perfect butler with no own opinion.
Not tiring ? Yes, definitely. I heard so many of this very clean sounding amps, which tired me to death (like Krells, ML, etc. - sorry folks).
Bass shy ? I thought that at first glance but after putting on records with lots of real bass I could not confirm it. Bass is maybe a bit more right ? My inverted chip-amp can play really loud (24V secondaries), and, although running hot then, shows no signs of giving up, and even the bass is very stable. This fact amazes me most, because the PSU is a slap in the face of a normal textbook one.
I already tried a 2 x 2200yF version (non-inverted), which sounded quite similar, but much more in the usual hifi sense. It is hard to describe: I just did not like that much than the 1000yf version (there were other slight differences too, I must admit, like metal films, FB ratio 20/1 and a NFB CD blocking cap).
The musical flow is there and well preserved but the presentation of details is quite different. But this is still just a test and not a defintive setup. I will try different input caps, an impedance smoothing resistor (.22 Ohm, like suggested), and other mods. But this amp is much "faster" than many others, the sound starts and stops in an breathtaking both secure and lively way, It is truly amazing.
I can confirm your observations about the sensitivness of the inverted version to mains noise. Although my amp is as quiet as its brothers (or sisters) and completetly noise and hum-free, it picks up mains noise quite heavily like the switching noise of the *******ing fridge from the kitchen nearby.
I read in another thread here on this forum about a very possible reason: In the inverted version the input (output of the source and cable) is a part of the feedback loop and therefore very sensitiv to picking up all kind of noise. In every mixing desk the most sensitive point is the input of the mixing amp connected to the mixing rails (usually a virtual ground inverted amplifier). Here normally every possible care is taken to sreen everything properly. Very expensive mixing desks therfore even have symmetric (balanced) mixing rails to get rid of the least bit of RFI noise picked up at this parts of the circuit.
Thor/Kwei was suggesting here (or elsewhere) to ground the resistor (or the short maybe) to the input ground first, not to the more noisy star ground between the psu caps. I did not do that, I soldered everything to the starground - I mean the whole circuit is about a square centimeter or less, so I did not see much benefit form that. But maybe ?
A solution, already suggested in this thread would be an input buffer outside of the feedback loop or a well designed preamp.
(That is the path the Final amps are following IMO. Did somebody try to reengineer these ones already ? The main thing seems to be, besides the odd dry battery power psu, a low gain preamp with low Z output, and an inverted LM1875 circuit with a two-part feedback resistor, one paralleled with a pot.)
Klaus
RFI
Alright, so I put together another inverted lm3875, bare bones variety: input resistor/cap, power supply caps, NFB resistor, log taper pot, starground. I notice that when I turn the volume all the way down, I can hear a radio station (barely, but its there).. Assuming I put my amp in well shielded box, do you think that will solve my RF problems? ( i was thinking a two layer steel copper case) Would the input/output/power supply wires undo my shielding efforts?
--Jordan
Alright, so I put together another inverted lm3875, bare bones variety: input resistor/cap, power supply caps, NFB resistor, log taper pot, starground. I notice that when I turn the volume all the way down, I can hear a radio station (barely, but its there).. Assuming I put my amp in well shielded box, do you think that will solve my RF problems? ( i was thinking a two layer steel copper case) Would the input/output/power supply wires undo my shielding efforts?
--Jordan
Re: RFI
--Jordan
Hrm, so I was poking around my amp, when I noticed that touching/pressing my input cap resulted in some noise/static.. So I took some aluminum foil, wrapped it gently around my cap, and tied the aluminum foil to ground.. Lo and behold, the RF noise was reduced dramaticially (to almost nothing). Very weird.. I guess my input cap was picking up most of the RFI? Also, I'm currently using 2uF 10% input cap.. Should I maybe use a larger value?JordanG said:
--Jordan
Philip,
I think it is unfair to compare the passive gc to another amp driven with an active pre. In my experience with driving the gc passively (with an external passive pre) and actively with a balanced line stage the active will make a huge improvement with respect to dynamics and bass.It sounds totally different and any shortcomings in the bass department are greatly reduced.
I think it is unfair to compare the passive gc to another amp driven with an active pre. In my experience with driving the gc passively (with an external passive pre) and actively with a balanced line stage the active will make a huge improvement with respect to dynamics and bass.It sounds totally different and any shortcomings in the bass department are greatly reduced.
Re: Re: RFI
A smaller value will be better from an RFI point of view. What is important here is that the physical size should be as small as possible and wiring as short as possible. Also, run the shielded cable as close to the cap's body as possible, if necessary cutting of excess cap leads.
Jan Didden
JordanG said:
A smaller value will be better from an RFI point of view. What is important here is that the physical size should be as small as possible and wiring as short as possible. Also, run the shielded cable as close to the cap's body as possible, if necessary cutting of excess cap leads.
Jan Didden
Protos,
Your right about the comparing "apples and oranges". I guess I didn't make my post clear that I am really very happy with this minimalist amp. I am going to pull out the pot to and see run the amp with my pre to see how it performs. I am also looking at a few other combinations including using a low voltage opamp for a pre amp and then into a LMxxxx/OPAxxx/TDAxxxx . I just need to iron out the power as the low value opamps use a lower drive voltage. TI has a brief white paper on the subject that has got me intersted. I am really just fiddling until all my aleph x monoblock parts get in and I'll start on that.
Your right about the comparing "apples and oranges". I guess I didn't make my post clear that I am really very happy with this minimalist amp. I am going to pull out the pot to and see run the amp with my pre to see how it performs. I am also looking at a few other combinations including using a low voltage opamp for a pre amp and then into a LMxxxx/OPAxxx/TDAxxxx . I just need to iron out the power as the low value opamps use a lower drive voltage. TI has a brief white paper on the subject that has got me intersted. I am really just fiddling until all my aleph x monoblock parts get in and I'll start on that.
Some interesting considerations on inverting op-amps ( advantages over non-inverting, feedback path length, using buffers etc.)
http://members.ozemail.com.au/~joeras/tubes_&__the_gainclones.htm
argo
http://members.ozemail.com.au/~joeras/tubes_&__the_gainclones.htm
argo
Yesterday's little project.
I had a few samples of the LM3875 and decided that I would give them a try. I built the circuit in the following pictures according to the generic amplifier seen in figure one of the National Semiconductor's specifications sheet. I have yet to introduce a music signal but I have tested it with standard test signals. I do not have any tabulated results however, into a 5-Ohm load it was producing ~30 W on 25 V rails before my supplies would go into over-current protection.
I will be continuing to develop the project as soon as I finish a pair of speakers. I will be adding an input coupling capacitor and experimenting with a few other modifications as is the spirit of this board. Capacitors are all Panasonic and resistors are all BC Components. Nothing special really.
Here it goes:
I had a few samples of the LM3875 and decided that I would give them a try. I built the circuit in the following pictures according to the generic amplifier seen in figure one of the National Semiconductor's specifications sheet. I have yet to introduce a music signal but I have tested it with standard test signals. I do not have any tabulated results however, into a 5-Ohm load it was producing ~30 W on 25 V rails before my supplies would go into over-current protection.
I will be continuing to develop the project as soon as I finish a pair of speakers. I will be adding an input coupling capacitor and experimenting with a few other modifications as is the spirit of this board. Capacitors are all Panasonic and resistors are all BC Components. Nothing special really.
Here it goes:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.