Measure the resistance of them cold. Has that value been taken into account in the design of the crossover ?
Hi everyone, a small update on my quest to hi-end audio 
First, I decided to leave the speaker crossovers alone as ultimately I will be bypassing it anyway. Maybe I'll replace the electrolytes with something half-decent and that's it for now.
Second, I have made a few improvements to my current GC:
- soldered proper speaker binding posts in place of previous temporary connectors (slight improvement in clarity)
- rearranged the cabling so that it's neat and power cables are away from signal cables (certain fuzziness is gone sound is definitely more ordered and layered)
- replaced the two 10uF Panasonic FC caps in PSU section with two 4.7uF Black Gate N that I found in one of my old projects (definite improvement, more bass, still extremely accurate; more focus, more "edge" in the sound; everything that is great in Gainclones but more)
- upgraded to tri-wiring. I added some no-name 1.5mm OFC cables that were left over from my HT setup. (on first impression - no audible improvement, perhaps even a sliver of transparency missing? I haven't had the time to audition this properly)
Overall, I'm very surprised by how much of a difference all these little things made. Especially the switch over to Black Gates. God, I wish I had the money to buy the big 1000uF ones a few years ago when they were still available! If the big ones bring similar qualities to the sound as the little ones do, then I WANT MORE!
Third, I have done some research, and got a few quotes, regarding my future amp. And it looks like I have a serious problem...
First, bridged configuration sounds slightly better in general, than a single-chip; on the other hand paralleling is somewhat lacking sound-wise.
Unfortunately, bridging amps means the speaker impedance is effectively halved, and that works out at 3Ohm for my 6Ohm speakers - not great. I could use LM4870 (which is basically two LM3886s packaged in a single-chip), as it has more current capability than LM3875 (and would make implementation of bridging simpler). I suppose I could also use slightly lower voltage than I had initially planned; something like 18 or 20VAC perhaps.
Another issue is, two stereo channels times three frequency channels, times two "bridges" - that's a twelve channel amp(!!) we're talking about. I'm not sure if this is still within my "simplicity principle" anymore...
Yet another problem is that bridged amps would need balanced signal. Which is not such a huge deal for the active crossovering setup because the Behringer xover has balanced outs. But this does severly limit flexibility of such an amp as a standalone "power amp".
Please share your thoughts on this.
First, I decided to leave the speaker crossovers alone as ultimately I will be bypassing it anyway. Maybe I'll replace the electrolytes with something half-decent and that's it for now.
Second, I have made a few improvements to my current GC:
- soldered proper speaker binding posts in place of previous temporary connectors (slight improvement in clarity)
- rearranged the cabling so that it's neat and power cables are away from signal cables (certain fuzziness is gone sound is definitely more ordered and layered)
- replaced the two 10uF Panasonic FC caps in PSU section with two 4.7uF Black Gate N that I found in one of my old projects (definite improvement, more bass, still extremely accurate; more focus, more "edge" in the sound; everything that is great in Gainclones but more)
- upgraded to tri-wiring. I added some no-name 1.5mm OFC cables that were left over from my HT setup. (on first impression - no audible improvement, perhaps even a sliver of transparency missing? I haven't had the time to audition this properly)
Overall, I'm very surprised by how much of a difference all these little things made. Especially the switch over to Black Gates. God, I wish I had the money to buy the big 1000uF ones a few years ago when they were still available! If the big ones bring similar qualities to the sound as the little ones do, then I WANT MORE!
Third, I have done some research, and got a few quotes, regarding my future amp. And it looks like I have a serious problem...
First, bridged configuration sounds slightly better in general, than a single-chip; on the other hand paralleling is somewhat lacking sound-wise.
Unfortunately, bridging amps means the speaker impedance is effectively halved, and that works out at 3Ohm for my 6Ohm speakers - not great. I could use LM4870 (which is basically two LM3886s packaged in a single-chip), as it has more current capability than LM3875 (and would make implementation of bridging simpler). I suppose I could also use slightly lower voltage than I had initially planned; something like 18 or 20VAC perhaps.
Another issue is, two stereo channels times three frequency channels, times two "bridges" - that's a twelve channel amp(!!) we're talking about. I'm not sure if this is still within my "simplicity principle" anymore...
Yet another problem is that bridged amps would need balanced signal. Which is not such a huge deal for the active crossovering setup because the Behringer xover has balanced outs. But this does severly limit flexibility of such an amp as a standalone "power amp".
Please share your thoughts on this.
Last edited:
And it also has a lower heatsinking capability which is especially bad for a bridged amplifier.
There are several ways to implement a bridged amp.
- use one half of the bridge in inverted configuration and you can use an unbalanced signal.
- use an inverting opamp to feed one half of the non-inverted bridge which has the same effect.
- buy (or build) a DI box. Passive versions are available from 10 € upward, active version from 25 € upward.
- buy (or build) an adapter that allows you to connect an unbalanced signal to a balanced input. Schematics are all over the Forum and the Web.
I think that your reduction in sound with paralleled amps may well be sub-optimal implementation rather than a problem with parallelling itself. Chipamps main weaknesses are lack of output current and increased distortion with lower impedance loads, both of which are fixed by paralleling. I'm working on a design at present with 8 X LM1876 (individually rather limited in output current) in bridged/parallel - definitely nothing lacking in the sound quality department but the individual amp gains do need well matching.
Precisely, which is why bridging and paralleling belong together. To get all the advantages of bridging without the drawbacks of higher distortion, paralleling becomes a necessity.
That is of course a possibility... But Peter Daniel told me that he never really liked the sound of paralleled chips either, and I imagine he would not have made many mistakes along the way... On the other hand, he did prefer the sound of bridged over single-chip.
Another thing is, if I do bridged x paralleled that will work out to the total of 24 channels, which is just plain ridiculous. That would probably defeat my purpose of creating something "simple and elegant", and it would definitely go over my budget.
I would rather put up with bass clipping on very high loudnesses, or accept slightly worse dynamics and just go the single-chip way.
At the minute I'm leaning towards bridged LM4870 running on low voltage.
I know, but from my (vast) experience with computer heatsinks, there is no heat problem that a big enough piece of copper won't resolve
I was actually considering trying water-cooling (my specialty) on this amp.Uncle,
if you use a simple single chipamp per speaker driver and design for near 60W of available maximum power to each driver you will have a system that is capable of creating music SPLs that are far greater than any 60W amplifier could ever put into a passive crossovered multidriver speaker.
I would guess that with direct driving and active filters, you can increase the maximum peak SPLs by at least +3dB and maybe approaching +8dB into a 4way speaker.
+3dB would be equivalent to 120W which would require bridged parallel set up.
+8dB would be equivalent to 370W which cannot be got from chipamps.
if you use a simple single chipamp per speaker driver and design for near 60W of available maximum power to each driver you will have a system that is capable of creating music SPLs that are far greater than any 60W amplifier could ever put into a passive crossovered multidriver speaker.
I would guess that with direct driving and active filters, you can increase the maximum peak SPLs by at least +3dB and maybe approaching +8dB into a 4way speaker.
+3dB would be equivalent to 120W which would require bridged parallel set up.
+8dB would be equivalent to 370W which cannot be got from chipamps.
I know what you meant, but I want bridged because it sounds better - not to get more SPL.
What I was saying in my previous post was that bridging is tricky with low-Ohm speakers, but I'm hoping that with heavy heatsinks and reasonable listening levels, and lowered voltage, AND the extra sensitivity from direct driving - the amps will work just fine.
I hope I'm making myself clear now
What I was saying in my previous post was that bridging is tricky with low-Ohm speakers, but I'm hoping that with heavy heatsinks and reasonable listening levels, and lowered voltage, AND the extra sensitivity from direct driving - the amps will work just fine.
I hope I'm making myself clear now
So then, my next question would be this. Does paralleling bridged amps reduce the sound quality below the level of a single, unbridged chip?
Ridiculous in hifi is just a matter of degree. Ever looked inside a Mark Levinson No.33 ?
Where I live, 5pin chip amps are about 3 to the USD. Hardly 2nd mortgage stuff even with 48 channels.
At the minute I'm leaning towards bridged LM4870 running on low voltage.
I know, but from my (vast) experience with computer heatsinks, there is no heat problem that a big enough piece of copper won't resolve
Unless you're doing sinewave testing or listen to massively compressed recordings at very high levels, I doubt very much that heatsinking will be a problem in an active set-up. Water cooling sounds like a sledgehammer to a hazelnut...
Hello,
Not sure where you’re at on your design criteria, but it seems obvious from your list of parts, that you intend to build a serious amplifier. If you're somewhat lacking in assembly equipment, I suggest staying with thru-hole components, and keep the design simple and discrete. IC designed amplifiers are neat, but can become unstable at the drop of a hat. Also damn near impossible to fabricate, without having PC board layout experience, and someone to cook your boards.
Do you use LT spice for simulation? If not, what Spice are you using? I have a number of very competent discrete designs, and would be willing to share some ideas, if you are interested.
Best regards,
Phoenix/Cauldron Amplifiers
Not sure where you’re at on your design criteria, but it seems obvious from your list of parts, that you intend to build a serious amplifier. If you're somewhat lacking in assembly equipment, I suggest staying with thru-hole components, and keep the design simple and discrete. IC designed amplifiers are neat, but can become unstable at the drop of a hat. Also damn near impossible to fabricate, without having PC board layout experience, and someone to cook your boards.
Do you use LT spice for simulation? If not, what Spice are you using? I have a number of very competent discrete designs, and would be willing to share some ideas, if you are interested.
Best regards,
Phoenix/Cauldron Amplifiers
You will probably get away with bridging only the woofer channels. During the process you can check whether bridged really sounds better than single IC.
It most certainly can. I lashed up a pair of 450W multiple parallel/bridged TDA7294 monoblocks many years ago. From memory, it had 16 chipamps per channel, 8 in parallel either side of the bridge. Admittedly it was 450W into 4R, 60V peak output. Nowadays, using TDA7293s would allow the envelope to be pushed further.
Hi,
LM3886 can run on 35V rails without problem. Given enough bridged parallel units this allows 64V peak or 45V RMS into any funloving load. That is 250W into 8 Ohm, 500W into 4 Ohm and 1000W into 2 Ohm.
Now 1000W into 2 Ohm would need 8pcs per channel purely on current limiting issues and probably 12 to 16pcs to handle the dissipation (8pcs would probably suffice for music though). But it CAN be done (which is not to say it should, mind you, or that I would suggest such an Amp).
Ciao T
LM3886 can run on 35V rails without problem. Given enough bridged parallel units this allows 64V peak or 45V RMS into any funloving load. That is 250W into 8 Ohm, 500W into 4 Ohm and 1000W into 2 Ohm.
Now 1000W into 2 Ohm would need 8pcs per channel purely on current limiting issues and probably 12 to 16pcs to handle the dissipation (8pcs would probably suffice for music though). But it CAN be done (which is not to say it should, mind you, or that I would suggest such an Amp).
Ciao T
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.