Hi! Im pretty new around here and would like some advise before i start my first amplifier (chipamp kit) build.
I have four non inverting lm3886 kits from "chipamp", four power supplies, two 350va torroidal transformers, and one hifi2000 pessante dissipante 2u case.
I have neutrik rca and xlr connectors, And all the requisite materials to ground the chassis and supply wall voltage.
These are the kits i have:
Non-Inverting LM3886 Dual Mono Kit | Chipamp Electronics
http://chipamp.com/beta2/wp-content/uploads/2013/12/lm3886-manual.pdf
Questions:
what is/are the best way to achieve balanced input (and ideally, realize the benefit of differential noise cancellation) with the chipamp, or must i drive it with rca? I will have the amp placed between the speakers, and the preamp some distance away, near my seating position - so balanced connection is ideal.
1) Is it possible to achieve a balanced connection, and how?
2) Is it possible to use four of these amp boards to safely drive one stereo pair of speakers, and how would i best achieve this?
3) should i bridge them into two mono amps, if so, how?
Or Should i use one each to drive + and - each side? (Would this double the gain? Is that too much?)
*If you can provide advice with explanations and drawings or photos, it would be really great.*
I really appreciate your input and advise!
Andrew
Background info....
The amp -may- be used to drive speakers down to 4ohms nominal impedance, and -potentially- martin logan electrostatics speakers in the future. It will definitely drive speakers at 8ohms nominal impedance. Safety, Stability and reliability are of concern and i seek advise to avoid making expensive mistakes with my tinkering.
Im not an engineer or technician - just a music enthusiast trying to get more involved in the hobby. Ive built the pearl2 and am currently finishing/debugging an aleph p1.7 preamp. I can solder and follow directions, but am not educated enough in audio design to make unassisted decisions.
I have four non inverting lm3886 kits from "chipamp", four power supplies, two 350va torroidal transformers, and one hifi2000 pessante dissipante 2u case.
I have neutrik rca and xlr connectors, And all the requisite materials to ground the chassis and supply wall voltage.
These are the kits i have:
Non-Inverting LM3886 Dual Mono Kit | Chipamp Electronics
http://chipamp.com/beta2/wp-content/uploads/2013/12/lm3886-manual.pdf
Questions:
what is/are the best way to achieve balanced input (and ideally, realize the benefit of differential noise cancellation) with the chipamp, or must i drive it with rca? I will have the amp placed between the speakers, and the preamp some distance away, near my seating position - so balanced connection is ideal.
1) Is it possible to achieve a balanced connection, and how?
2) Is it possible to use four of these amp boards to safely drive one stereo pair of speakers, and how would i best achieve this?
3) should i bridge them into two mono amps, if so, how?
Or Should i use one each to drive + and - each side? (Would this double the gain? Is that too much?)
*If you can provide advice with explanations and drawings or photos, it would be really great.*
I really appreciate your input and advise!
Andrew
Background info....
The amp -may- be used to drive speakers down to 4ohms nominal impedance, and -potentially- martin logan electrostatics speakers in the future. It will definitely drive speakers at 8ohms nominal impedance. Safety, Stability and reliability are of concern and i seek advise to avoid making expensive mistakes with my tinkering.
Im not an engineer or technician - just a music enthusiast trying to get more involved in the hobby. Ive built the pearl2 and am currently finishing/debugging an aleph p1.7 preamp. I can solder and follow directions, but am not educated enough in audio design to make unassisted decisions.
Last edited:
0 and 1) Check out DIYA's grounding and interconnect article and its references, particularly the Bill Whitlock bits. All op amps are differential amplifiers so, Nigel, I assume you're referring to what's more commonly referred to as a difference amplifier (or, less frequently, a subtractive amplifier). However, it's unclear what amount of CMRR and imbalance tolerance you've in mind.
2 and 3) I think the OP's mainly looking for TI app note AN-1192, though the poll over in the multiway forum about how much voltage one's speakers require is likely useful in illustrating one LM3886 is usually more than enough. Nigel, can you be more specific as to the safety differences? The faults springing to mind in bridged or parallel are all seem of more or less equal severity. Usually the simplest protection is choosing the lowest suitable rails as one has to kind of work at it to get the 3886's 11.5A current limiting to trip in normal operation.
2 and 3) I think the OP's mainly looking for TI app note AN-1192, though the poll over in the multiway forum about how much voltage one's speakers require is likely useful in illustrating one LM3886 is usually more than enough. Nigel, can you be more specific as to the safety differences? The faults springing to mind in bridged or parallel are all seem of more or less equal severity. Usually the simplest protection is choosing the lowest suitable rails as one has to kind of work at it to get the 3886's 11.5A current limiting to trip in normal operation.
I'd suggest checking out the grounding article and references as well, then. AC coupling mostly just makes impedance matching for good CMRR harder (unless by AC coupling you mean a signal trafo rather than caps).
The standard solution for preventing parallel amplifier fights is low value coupling resistors---see Rout in PA100 and Ron in BPA200 in AN-1192---as inter or intra-die variability means no two amplifiers will track exactly no matter how tightly matched the feedback impedances. A better argument for bridging is probably odd order distortion cancellation, though there's some tradeoffs in output impedance between the two topologies.
The standard solution for preventing parallel amplifier fights is low value coupling resistors---see Rout in PA100 and Ron in BPA200 in AN-1192---as inter or intra-die variability means no two amplifiers will track exactly no matter how tightly matched the feedback impedances. A better argument for bridging is probably odd order distortion cancellation, though there's some tradeoffs in output impedance between the two topologies.
Thank you very much for your thoughtful responses! I''m going to dig into those recommended readings and see what I can come up with. It seems CLear to me that in order for me to achieve a parallel set up I will require boards that will allow careful balancing of dc offset and that will include small 0.1ohm series resistance to each. I'm not so sure how critical the dc offset adjustability will be as I've not build up the briangt chipamp boards yet but, I can't find anyone who has tried paralleling two individual boards. Probably not possible. All cases of parallel and bridge parallel amps I've seen are all on one board. I was curious if it was possible with separate one chip chipamp boards.
I've seen amps where a balanced input signal is split between two amps and I might try it... (Is this safe?)
However a safer bet is probably to just use an integrate balanced to se converter chip to enable balanced input and just use the amp as a four channel amp.
. I can still bridge them though if I need, right?
Any configuration can be matched. It's just that the more complex it is the more work it is to obtain a given level of matching. For example, pretty decent connectors have something like a 10 mOhm contact resistance, aging to 20 mOhm. For wiring from two boards to a third that's 40 to 80 mOhms to deal with, either by not having multiple boards or skipping the connectors and soldering directly.
I wouldn't bother with bridging or paralleling until it's established the extra power's actually needed. But if DC offsets aren't matched within a few millivolts then the fix is a better control loop, either in the form of a DC servo, correct Kelvin sense implementation, or something else. You can work through the balanced input split and bridge cases as you need them.
I wouldn't bother with bridging or paralleling until it's established the extra power's actually needed. But if DC offsets aren't matched within a few millivolts then the fix is a better control loop, either in the form of a DC servo, correct Kelvin sense implementation, or something else. You can work through the balanced input split and bridge cases as you need them.
Did you ever get it built? I am also interested in a balanced gainclone amp, but I need 3 channels with level adjustments.
If you do that...won't you be losing the balanced qualities, or is it only necessary before and after the amp?
Umm, no, post 12 is incorrect. The 3886 is a differential amplifier like any other. The above remarks demonstrate lack of understanding as to how to connect it as a difference amplifier. Or perhaps failure to grasp a differential amplifier's Vin and Vout need to share a reference.
Terminology aside, you can certainly add a differential input to your ChipAmp boards. One way to accomplish this is to use a THAT1200 as a differential-to-single ended driver. XLR goes into the THAT, output of THAT connects to LM3886 input.
As Twest points out, you can also rig the LM3886 as a difference amplifier. This will require some hacking of your circuit boards, but can be done. The main drawback of this solution is that its input impedance will be in the low kΩ range, which is tough to drive for many sources. There are ways to increase the input impedance. The most common is to essentially build a THAT1200 with op-amps.
If your goal is to be able to plug in a differential XLR connection to your LM3886, I'd go with the THAT1200.
As far as driving electrostatic speakers and other capacitive loads go: Make sure you add a Thiele network in series with the output, otherwise you're guaranteed to have problems with instability. I find 2 uH || 1.5 Ω works well. The Zobel network (C+R to ground) is already included on the ChipAmp boards.
Also note that when driving 4 Ω loads, I recommend keeping the supply voltage at or below ±28 V. This ensures that the power dissipation in the chip is kept to a safe level (assuming a sizable external heat sink) and that the LM3886 won't exceed its 7 A (min) output current spec.
Tom
As Twest points out, you can also rig the LM3886 as a difference amplifier. This will require some hacking of your circuit boards, but can be done. The main drawback of this solution is that its input impedance will be in the low kΩ range, which is tough to drive for many sources. There are ways to increase the input impedance. The most common is to essentially build a THAT1200 with op-amps.
If your goal is to be able to plug in a differential XLR connection to your LM3886, I'd go with the THAT1200.
As far as driving electrostatic speakers and other capacitive loads go: Make sure you add a Thiele network in series with the output, otherwise you're guaranteed to have problems with instability. I find 2 uH || 1.5 Ω works well. The Zobel network (C+R to ground) is already included on the ChipAmp boards.
Also note that when driving 4 Ω loads, I recommend keeping the supply voltage at or below ±28 V. This ensures that the power dissipation in the chip is kept to a safe level (assuming a sizable external heat sink) and that the LM3886 won't exceed its 7 A (min) output current spec.
Tom
Last edited:
Eh, take the time to look into the differences and I'll suspect you'll be surprised.
I notice I did make an unfortunate typo. That should be "Vin and Vout need not share a reference."
Actually, considering the number of Modulus and Parallel boards sold since I put you on to the 1200, I suspect the most common approach in the DIY space is to use a 1200.
Hypex almost certainly has a larger installed base of LME49720 front end instrumentation amplifier topologies, though that's not relevant to a 3886 thread.
Last edited:
Actually, considering the number of Modulus and Parallel boards sold since I put you on to the 1200, I suspect the most common approach in the DIY space is to use a 1200.
That certainly would be my recommendation. No reason to reinvent the wheel. In particular as it's pretty darn hard to get the CMRR of the THAT1200 in a op-amp+resistors based solution.
Tom
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.