Thanks !
By looking at the schematic, I've observed that the following block is repeating itself :
1Meg ohms
0.22 ohms
22k ohms
3.3uF
LM3886
If I understand correctly, to increase or decrease the power output of an op-amp based amplifier, we simply need to add or remove "blocks" defined previously.
I guess that there's some limitations such as
we can't add a huge number of op-amp because of their current limitation. Right ?
Can we use any op-amp using this topology or the LM3886 has something special in it that makes it more adapted for this task compared to others ?
thanks !
By looking at the schematic, I've observed that the following block is repeating itself :
1Meg ohms
0.22 ohms
22k ohms
3.3uF
LM3886
If I understand correctly, to increase or decrease the power output of an op-amp based amplifier, we simply need to add or remove "blocks" defined previously.
I guess that there's some limitations such as
we can't add a huge number of op-amp because of their current limitation. Right ?
Can we use any op-amp using this topology or the LM3886 has something special in it that makes it more adapted for this task compared to others ?
thanks !
Offset
hi Joe
I am having a bit of a problem with a single pot for every three chips; surely you can't null the offset if you don't employ separate pots per chip.
Why the large value for the FB resistor? Is so much gain really essential?
This ECC88 follower still looks a bit out of place. A big contrast to the high PSRR of the chips. I understand that a circuit like this can add second harmonics and give more body and 'pleasantness' to the sound but it's probably a good idea to apply just a bit of extra engineering. How about a current source at least?
cheers
peter
hi Joe
I am having a bit of a problem with a single pot for every three chips; surely you can't null the offset if you don't employ separate pots per chip.
Why the large value for the FB resistor? Is so much gain really essential?
This ECC88 follower still looks a bit out of place. A big contrast to the high PSRR of the chips. I understand that a circuit like this can add second harmonics and give more body and 'pleasantness' to the sound but it's probably a good idea to apply just a bit of extra engineering. How about a current source at least?
cheers
peter
fdegrove said:Hi,
The idea of balancing the input is quite appealing.
It wouldn't require a great deal of money to do so and would be quite interesting to compare to the SE version.
I had the idea of putting the buffer stage in a separate case so it could easily be preceded by a source selector and volctrl.
It could than be used to drive longer lengths of IC without perceivable attenuation at HF and you could build the amps as monoblocks and put them closer to the LS.
Is this a sound (no pun intended) concept?
Cheers,
It can be done that way, as I see you describing it. The main thing is that the Tx see a Lo-Z source, indeed this always gives the lowest distortion with line Txs.
The only objection is the feedback path will still extend back to the source and if you remote site it (the buffer) then it does extend through the interconnects. But I won't be dogmatic on that score, the Lo-Z drive takes precedence.
JR
Elkaid said:
Me thinks you answered your own question.
fdegrove said:Hi,
Yes, but with opposed polarity so it's a phase splitter I suppose.
It serves other clever purposes but I leave to up to J.R.
Cheers,
Thank you, most gracious.
Indeed you guys have it right, it is both. I was comtemplating doing this differentially, but doing this with low or near unity gain, not easy but can be done, would likely give pour CMRR. I discussed this with Allen Wright on the blower (oops, aussie for phone) and he was totally against it. So I got the brainstorm to do it using a 1:1 Tx. It then also occured to me that Rowland praised the Jensen line Tx on his web site in one of his online 'papers.' He praises the high CMRR. Bingo! I thought.
JR
Hi,
Taking the interconnect in the FB loop may have its' advantages as well.
In my system I take the LS and its cable in the overall NFB loop and I use the topology as I described it for the linepreamp.
Works fine for me and I think it's just common sense, YMMV.
It doesn't cost an arm and legg to try it out and it has the added advantage of current driving the cables iso of voltage drive.
Cheers,
Taking the interconnect in the FB loop may have its' advantages as well.
In my system I take the LS and its cable in the overall NFB loop and I use the topology as I described it for the linepreamp.
Works fine for me and I think it's just common sense, YMMV.
It doesn't cost an arm and legg to try it out and it has the added advantage of current driving the cables iso of voltage drive.
Cheers,
Elkaid said:
Oh boy! I could spend the next two hours on how I got the above values, not just individually as they do inter-act so much. There is a question of target gain, the input impedance seen by the buffer that in turn affects the value of the coupling caps, which I wanted kept at 3.3uF as per earlier Hybrid IGC. The inline balanced Lo-Pass filter also has a part, a useful one, to play.
I used a simulation program to get all these balanced just right. In the end the two main feedback values ended up very similar to before, but the 22K are now in parallel (22/3) but in series with the block of the same on the other side. Hence adding more 'blocks' makes it harder to drive. The 4K7 balances out the gain and also gives the means to provide Lo-Pass and 'ups' the drive impedance, which in the end is 20K48, very close to the 'single blocks.' So 3.3uF can be used in all situs. Total gain 29dB approx.
The LM3886 is needed as it has more grunt than the 3876, even though in terms of rated Watts it's not much higher. The current rating is 75% higher.
JR
Re: Offset
You can do that. With the 1M feedback resistor the individual offsets have been less than 10mV in 90% of the cases. So a single resistor at 333K333 should give very low offset, just trim it a little if not.
29dB. I usually aim for a gain of 20 times, or 26dB. This amp will be used with a voltage sensitivity 6-7 dB lower than what I use. The gain should be just about OK.
I would fit a CS. I am not TELLING YOU how to build amps. This is not a kit instruction. This is for serious DIY'ourselfers. Not only would I fit a CS but also a bootstrapped constant voltage device above the cathode follower... but that is another topic. This is a Super Linear Cathode Follower has VERY low distortion and is what I use all the time. Yes, I am keeping a few things up my sleeve on the commercial JLTi amp that I manufacture.
JR
analog_sa said:
You can do that. With the 1M feedback resistor the individual offsets have been less than 10mV in 90% of the cases. So a single resistor at 333K333 should give very low offset, just trim it a little if not.
29dB. I usually aim for a gain of 20 times, or 26dB. This amp will be used with a voltage sensitivity 6-7 dB lower than what I use. The gain should be just about OK.
I would fit a CS. I am not TELLING YOU how to build amps. This is not a kit instruction. This is for serious DIY'ourselfers. Not only would I fit a CS but also a bootstrapped constant voltage device above the cathode follower... but that is another topic. This is a Super Linear Cathode Follower has VERY low distortion and is what I use all the time. Yes, I am keeping a few things up my sleeve on the commercial JLTi amp that I manufacture.
JR
Elkaid said:
Not just a thought, it could be done but would it be wise?
IF the standard single GC with STIFF power supply is capable of 50W (I have measured 20.5V RMS @ 8R which is more than 50W), then a balanced version will be 200W into 8R, all other things being equal (not always true but for simplicity's sake not far from the end result), and who knows what into 4 Ohm?
Since 25V AC gives around 35V DC, then the next easily available is value in TXs is 18V AC, and that should give us near 25V. So that was kept in mind when specifying 25V.
The bottom line is this: When we go balanced we double our potential voltage swing, which also doubles current into the same load, meaning four times the power. So if we don't want that much more power we have to back off on the rails.
That's what I've done.
It also gives us a better SOA (Safe Operating Area) for driving lower impedances. Better muscle characteristics.
JR
Hi,
What I think Elkaid had in mind is raising the B+ on the semis so they could share the same supply voltage withe the CF.
Balanced or not, doesn't really matter here.
BTW, are you sure a balanced design actually would double output power?
I am certainly no expert on semiconductors but in a valved design that one wouldn't fly for sure...
Leave you to it,
What I think Elkaid had in mind is raising the B+ on the semis so they could share the same supply voltage withe the CF.
Balanced or not, doesn't really matter here.
BTW, are you sure a balanced design actually would double output power?
I am certainly no expert on semiconductors but in a valved design that one wouldn't fly for sure...
Leave you to it,
Hi,
I read you, but I wonder how that one could be pulled off?
If we look at the outputstage which is a PP design already I fail to see where on earth the quadrupling of power output can come from...
After all, the balancing i.e making symmetrical, only requires another buffer at the input...
Raising the voltage of the output stages is not changing topology, it will only put the whole cicruit on the brink of stability, it may give you some more Watts but...
Cheers,
I read you, but I wonder how that one could be pulled off?
If we look at the outputstage which is a PP design already I fail to see where on earth the quadrupling of power output can come from...
After all, the balancing i.e making symmetrical, only requires another buffer at the input...
Raising the voltage of the output stages is not changing topology, it will only put the whole cicruit on the brink of stability, it may give you some more Watts but...
Cheers,
One way of looking at it is that it is two amplifiers, each driving a load of one-half the impedance of the real load.fdegrove said:Hi,
I read you, but I wonder how that one could be pulled off?
If we look at the outputstage which is a PP design already I fail to see where on earth the quadrupling of power output can come from...
After all, the balancing i.e making symmetrical, only requires another buffer at the input...
Raising the voltage of the output stages is not changing topology, it will only put the whole cicruit on the brink of stability, it may give you some more Watts but...
Cheers,
Each of the amplifiers will be driving the same voltage into half the impedance, so EACH amplifier will put out twice the power. The two amplifiers together will put out 4x the power.
That assumes that the amplifiers can handle the additional current requirements (i.e. double).
Hi,
Thanks for your reply Paul, but:
Keeping in mind this condition, I wonder how:
This is going to pull off the miraculous quadrupling of the power output?
Have I missed something inbetween?
The outputstage is balanced as it is, that's what the xformer is there for: it's a phasesplitter anyway so if we want a balanced design we only have to look at the input to deal with that...now, my question is, how on earth is that going to yield a fourfold/double in power output?
That's the part I do not get, you see?
Cheers,
Thanks for your reply Paul, but:
Keeping in mind this condition, I wonder how:
This is going to pull off the miraculous quadrupling of the power output?
Have I missed something inbetween?
The outputstage is balanced as it is, that's what the xformer is there for: it's a phasesplitter anyway so if we want a balanced design we only have to look at the input to deal with that...now, my question is, how on earth is that going to yield a fourfold/double in power output?
That's the part I do not get, you see?
Cheers,
It's true that the output power gets four times larger when you double the output swing. But that doesn't mean you get four times as much output power with two chips. The poor chips would blow themselves up
The maximum output voltage of the chips will be reduced, probably by the maximum of current deliverable by the chips before the protection kicks in.
Fedde
The maximum output voltage of the chips will be reduced, probably by the maximum of current deliverable by the chips before the protection kicks in.
Fedde
Hi Frank
The quadrupling of output power in a bridged amp is obvious due to the the doubling of available ourput voltage. It is pretty much a free lunch if your output devices and power supply can take the current, if not you simply add output devices. Even if the particlular devices are incapable of supplying the current you still get a great improvement in subjective headroom, partially due to the higher voltage and partially, i think, to the decoupling of load currents from both signal and PS ground.
cheers
peter
The quadrupling of output power in a bridged amp is obvious due to the the doubling of available ourput voltage. It is pretty much a free lunch if your output devices and power supply can take the current, if not you simply add output devices. Even if the particlular devices are incapable of supplying the current you still get a great improvement in subjective headroom, partially due to the higher voltage and partially, i think, to the decoupling of load currents from both signal and PS ground.
cheers
peter
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.