I have 4 LM3886 ic's sitting doing nothing and really want an amp with a bit more weight behind it than my single chipped lm3886 has. Ive been looking into bridged and parallel setups and have come to the conclusion that (maybe incorrectly) bridged is better as long as you take into account the extra heat etc.
Im going to use matrix board and semi's referance for the bridged circuit but what id like is to A, know what sort of freedom i have to experiment with caps in the audio circuit. And B if any one has a schimatic for a power supply i can use with this with a 25-0-25 trafo?
Im going to use matrix board and semi's referance for the bridged circuit but what id like is to A, know what sort of freedom i have to experiment with caps in the audio circuit. And B if any one has a schimatic for a power supply i can use with this with a 25-0-25 trafo?
http://www.diyaudio.com/forums/attachment.php?s=&postid=499002&stamp=1098741354
Any good for what i need?
Any good for what i need?
A 25-0-25 trafo will give you roughly 35V rails, quite a bit higher than is really required for bridge configuration. Unless you have 16 ohm speakers, that is. If not, expect SPiKE to kick in very, very frequently at high levels.
I built a parallel amp myself, and couldn't be happier. Using good quality resistors instead of generic (only in the feedback position though) really helps illuminate differences between rigs, sources and recordings. And having about 20 amps of current on tap helps drive some inefficient loads, and also gives the chipamp a lot more grunt.
That PSU will work, but with the caveat of a high rail voltage. Ideally 20V transformers should be used for 28V rails, for a bridge setup driving 8 ohm speakers. And you do need to take care of the heat.
I built a parallel amp myself, and couldn't be happier. Using good quality resistors instead of generic (only in the feedback position though) really helps illuminate differences between rigs, sources and recordings. And having about 20 amps of current on tap helps drive some inefficient loads, and also gives the chipamp a lot more grunt.
That PSU will work, but with the caveat of a high rail voltage. Ideally 20V transformers should be used for 28V rails, for a bridge setup driving 8 ohm speakers. And you do need to take care of the heat.
Re: Re: Bridged LM3886 psu.
I had a look at semis referance PSU and to be hounest i suppose i dont really understand it (im still learning especially were schimatics are concerned.
The only part i really understand is the two on the left and im unsure how these work together, any one care to elaborate the schimatic?
pacificblue said:
I had a look at semis referance PSU and to be hounest i suppose i dont really understand it (im still learning especially were schimatics are concerned.
The only part i really understand is the two on the left and im unsure how these work together, any one care to elaborate the schimatic?
This could help you to get a grip on power supplies.
After that you look at the Bill of Materials in AN-1192 and compare the designators with the schematic you posted. Things will hopefully become clearer then.
By the way, the upper two parts of that schematic are for the power amplifier, which is the interesting part for you. The lower three parts are for the op amps that are used as DC servos in the BPA-200, so probably not so important for you to start with.
After that you look at the Bill of Materials in AN-1192 and compare the designators with the schematic you posted. Things will hopefully become clearer then.
By the way, the upper two parts of that schematic are for the power amplifier, which is the interesting part for you. The lower three parts are for the op amps that are used as DC servos in the BPA-200, so probably not so important for you to start with.
Also, check out this spreadsheet to get an idea on power output, heat levels, etc. For bridged, put half the impedance of your speakers in the impedance field.
http://www.national.com/appinfo/audio/files/Overture_Design_Guide15.xls
http://www.national.com/appinfo/audio/files/Overture_Design_Guide15.xls
Hi,
bridged gives double the power into double the load impedance.
If you use +-35Vdc for your supply rails then each 3886 chipamp can give about 60W into 8r0.
When you bridge two 3886 chipamps off +-35Vdc supplies you should get 120W into 16r0, if the Power Supply is correctly sized to allow 120W at your chosen bandwidth.
bridged gives double the power into double the load impedance.
If you use +-35Vdc for your supply rails then each 3886 chipamp can give about 60W into 8r0.
When you bridge two 3886 chipamps off +-35Vdc supplies you should get 120W into 16r0, if the Power Supply is correctly sized to allow 120W at your chosen bandwidth.
Andrew, I think you're mixing it up with parallel. Parallel doubles the power and doubles the impedance (good for 4 ohm or lower loads). Bridged doubles the power and halves the impedance (good for 8 ohm or higher loads).
Two bridged LM3886s into an 8 ohm load: each LM3886 puts out the usual 60W or whatever, but into half the load (4 ohms). Also, as you are doubling the power (2 chips) and halving the impedance, you produce about 4 times as much heat.
Two bridged LM3886s into an 8 ohm load: each LM3886 puts out the usual 60W or whatever, but into half the load (4 ohms). Also, as you are doubling the power (2 chips) and halving the impedance, you produce about 4 times as much heat.
Hi Red,
no, what I said is correct.
You are looking at the effective impedance seen by each amplifier when in bridged configuration.
To take your viewpoint and using my data of 120W into 16r0; each amplifier delivers 60W into an effective 8r0. Note it is not double the power into half the impedance.
If you bridge AND halve the impedance you do not get four times the heat, you will more likely get a destroyed chipamp.
no, what I said is correct.
You are looking at the effective impedance seen by each amplifier when in bridged configuration.
To take your viewpoint and using my data of 120W into 16r0; each amplifier delivers 60W into an effective 8r0. Note it is not double the power into half the impedance.
If you bridge AND halve the impedance you do not get four times the heat, you will more likely get a destroyed chipamp.
I guess maybe we're trying to say the same thing. You're saying bridged 120W into 16 ohm is each amp putting 60W into 8 ohm. I'm saying bridged 120W into 8 ohm is each amp putting 60W into 4 ohm.
For what I'm talking about heat, imagine this. You have an 8 ohm speaker and a pair of bridged LM3886s set up for 60W each into 4 ohms. Running a single LM3886 into 4 ohms produces twice the heat of running the same setup into 8 ohms. Since you have 2 LM3886s, you have twice the heat again. That's 4 times the heat of a single LM3886 into an 8 ohm load. Not per chip, just total.
For what I'm talking about heat, imagine this. You have an 8 ohm speaker and a pair of bridged LM3886s set up for 60W each into 4 ohms. Running a single LM3886 into 4 ohms produces twice the heat of running the same setup into 8 ohms. Since you have 2 LM3886s, you have twice the heat again. That's 4 times the heat of a single LM3886 into an 8 ohm load. Not per chip, just total.
I don't know much about how you get the total impedance of a set of speakers, so I can't help much. From what I've read though, it's recommended strongly that for anything less than 8 ohm, you don't go bridged. You could go parallel, or bridged parallel (requiring 4 LM3886s per channel, but putting out 4 times the power of a single LM3886).
When in bridged mode, each amp sees half the actual impedance.
When in parallel mode, each amp sees twice the actual impedance.
In bridged mode, a 8 ohm load looks like 4 ohms to each chipamp.
In parallel mode, an 8 ohm load looks like 16 ohm to each chipamp.
At 35 volt rails, operation into an 8 ohm load in bridged mode is at the edge of safe area for a LM3886. Since each amp sees 4 ohms effective load, and the LM3886 is specified for 4 ohm load at 30V rails, I'm not sure it's a good idea to run such high rails. On the other hand, I've run chips out of spec on numerous occasions (not the National chips, they're too expensive for that) without failure.
It would be difficult to create a 16 ohm load, unless one were to connect drivers in series. If you did though, that would be a nifty way to get a lot of power out of these chips, as Andrew states. I don't know if they make commercial speakers with this high an impedance, though!
@kanifee: 8 ohm is not constant, but typical impedance. It is almost definite that the impedance will drop below that in some cases. A parallel amp helps the chips withstand high current demands without breaking a sweat. However it may not give you more 'punch' at lower volumes or at any volume.
If you drop the supplies to 30V rails (a 20V trafo works well), you can use a bridged aplifier, that will increase the power output while keeping within the safety margins of the chips.
Edit: A reference link: http://www.national.com/an/AN/AN-1192.pdf
Table 1 on page 4 summarises the supply voltages for different amp configurations and speaker impedances.
When in parallel mode, each amp sees twice the actual impedance.
In bridged mode, a 8 ohm load looks like 4 ohms to each chipamp.
In parallel mode, an 8 ohm load looks like 16 ohm to each chipamp.
At 35 volt rails, operation into an 8 ohm load in bridged mode is at the edge of safe area for a LM3886. Since each amp sees 4 ohms effective load, and the LM3886 is specified for 4 ohm load at 30V rails, I'm not sure it's a good idea to run such high rails. On the other hand, I've run chips out of spec on numerous occasions (not the National chips, they're too expensive for that) without failure.
It would be difficult to create a 16 ohm load, unless one were to connect drivers in series. If you did though, that would be a nifty way to get a lot of power out of these chips, as Andrew states. I don't know if they make commercial speakers with this high an impedance, though!
@kanifee: 8 ohm is not constant, but typical impedance. It is almost definite that the impedance will drop below that in some cases. A parallel amp helps the chips withstand high current demands without breaking a sweat. However it may not give you more 'punch' at lower volumes or at any volume.
If you drop the supplies to 30V rails (a 20V trafo works well), you can use a bridged aplifier, that will increase the power output while keeping within the safety margins of the chips.
Edit: A reference link: http://www.national.com/an/AN/AN-1192.pdf
Table 1 on page 4 summarises the supply voltages for different amp configurations and speaker impedances.
+-30Vdc into a chipamp driving a 4ohm reactive load is in my opinion asking too much of the 3886.
I would be looking @ <+-28Vdc and maybe even lower than +-25Vdc when at quiescent voltage.
Otherwise the chip runs out of current capacity to satisfy the transient demand on fast starting/stopping signals.
Similarly running the amps hotter when driving the lower impedance loads brings in the Spike protection sooner. In other words designing for 60W into 4r0 (+-25.5Vdc when loaded) will not give as good performance as designing for 60W into 8r0 (+-34Vdc when loaded).
Note, these loaded voltages are while the supply has sagged during the music signal peak and at the bottom of the ripple voltage between charging pulses from the mains. Both of these voltage losses (sag & ripple) are worse when driving a lower load impedance and in addition the voltage loss through the chip is also worse when on lower impedance.
I would be looking @ <+-28Vdc and maybe even lower than +-25Vdc when at quiescent voltage.
Otherwise the chip runs out of current capacity to satisfy the transient demand on fast starting/stopping signals.
Similarly running the amps hotter when driving the lower impedance loads brings in the Spike protection sooner. In other words designing for 60W into 4r0 (+-25.5Vdc when loaded) will not give as good performance as designing for 60W into 8r0 (+-34Vdc when loaded).
Note, these loaded voltages are while the supply has sagged during the music signal peak and at the bottom of the ripple voltage between charging pulses from the mains. Both of these voltage losses (sag & ripple) are worse when driving a lower load impedance and in addition the voltage loss through the chip is also worse when on lower impedance.
I'm building a bridged LM3886 for 8ohm speakers at the moment. I'm using LM4780s (just two LM3886s in one chip), and the Overture Design Guide tells me anything above 25V rails is going to need a fan. So I'm running approximately 25V (won't know for sure until I test the transformer). Although, separate LM3886s will dissipate more heat than two combined in one chip.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.