Hello,
I want to build two LM3886 based amp, I have the PCBs with regulator a caps on it, but do not have transformers yet. I do not need much power, for power I think +- 24V would be enough for these, but I do not know where does the LM3886 performs the best?
What is your experience? Should I shoot for near the max voltage because that is where it performs the best or is that the minimum rail voltage?
Boxes will be JPW Sonatas on both.
Also, how about switching power supplies. Did you guys tried with switching PSUs? (better ones, like Meanwell.)
Thanks,
JG
I want to build two LM3886 based amp, I have the PCBs with regulator a caps on it, but do not have transformers yet. I do not need much power, for power I think +- 24V would be enough for these, but I do not know where does the LM3886 performs the best?
What is your experience? Should I shoot for near the max voltage because that is where it performs the best or is that the minimum rail voltage?
Boxes will be JPW Sonatas on both.
Also, how about switching power supplies. Did you guys tried with switching PSUs? (better ones, like Meanwell.)
Thanks,
JG
-/+24 volts sounds good to me. You need an 18-0-18 volt transformer to provide that. Running at near the maximum is not recommended because the device dissipation increases dramatically. Higher is not always better.
SMPS are fine as long as you can get a suitably rated one that offers dual rail output. As Nigel says, SMPS does not need the high value of caps found in linear supplies.
You should be conversant with correct implementation of an SMPS but they are a great choice.
SMPS are fine as long as you can get a suitably rated one that offers dual rail output. As Nigel says, SMPS does not need the high value of caps found in linear supplies.
You should be conversant with correct implementation of an SMPS but they are a great choice.
The main problem with chip amps compared to discrete ones is the temperature variations of the outputs effecting the low level. For that purpose the LM3886 is optimized to function at constant dissipation at certain load/supply given at the table bellow.
If the load is 8 ohms you can see that with +/-25v you get near constant dissipation. The idea of Gaincard amplifier is to use higher but sagging supply due to low reservoir capacitors 2200uf so that at low power the disdipation equalizes to that of above 5w.
Normal chip amps have the output transistors heat up from 25°C to max. 150°C that is 125°C max increment, the LM3886 have unique in the history of semiconductors to use 250°C operating with protection, that is increment of 225°C instead of 125°C ordinary. This is not for decoration, the datasheet clearly gives heatsink size that makes the amp to function at 120°C junction temperature. This why the Gaincard amp uses tiny enclosure the size of CD casing for 2×25W and runs at untouchable temperature.
If the load is 8 ohms you can see that with +/-25v you get near constant dissipation. The idea of Gaincard amplifier is to use higher but sagging supply due to low reservoir capacitors 2200uf so that at low power the disdipation equalizes to that of above 5w.
Normal chip amps have the output transistors heat up from 25°C to max. 150°C that is 125°C max increment, the LM3886 have unique in the history of semiconductors to use 250°C operating with protection, that is increment of 225°C instead of 125°C ordinary. This is not for decoration, the datasheet clearly gives heatsink size that makes the amp to function at 120°C junction temperature. This why the Gaincard amp uses tiny enclosure the size of CD casing for 2×25W and runs at untouchable temperature.
Attachments
Last edited:
Because the LM3886 can survive high temperatures doesn't mean it is optimized to work at those temperatures. Datasheet specs are given for 25°C. The SPIKE app note makes clear that running the amp hot degrades its operation.
Furthermore, domestic amps operate usually most of the time in the vertical section of that dissipation vs output power curve. So good luck with constant dissipation.
+/-24Vdc on a nominal 8R load is fine.
Furthermore, domestic amps operate usually most of the time in the vertical section of that dissipation vs output power curve. So good luck with constant dissipation.
+/-24Vdc on a nominal 8R load is fine.
The LM3886 design published in Bob Cordell's book Designing Audio Power Amplifiers runs from ±40 volt DC supplies. However those rail voltages are uncomfortably close to the absolute maximum rating in the LM3886 datasheet (±42 volts) and I would recommend that inexperienced builders aim for a much less aggressive target, i.e., a much lower maximum supply voltage. This gives much greater safety margins, in case your transformer is out of spec, or your AC mains is out of spec, or Murphy's Law conspires against you.
Many intended-for-beginners LM3886 project designs, recommend using a 200VA power transformer with (2 x 22VACrms) secondaries. This gives about ±28 volts DC on the supply rails, IF your transformer is in spec and IF your AC mains is in spec. You can estimate the output power delivered to the loudspeakers, when the supply rails are ±28 volts DC, by studying Figure 37 of the LM3886 datasheet. You can also estimate the maximum heat dissipated by the LM3886 chip (and pulled away by an enormous heatsink conservatively chosen by YOU), when the supply rails are ±28 volts DC, by studying Figures 35 and 36 of the LM3886 datasheet.
_
Many intended-for-beginners LM3886 project designs, recommend using a 200VA power transformer with (2 x 22VACrms) secondaries. This gives about ±28 volts DC on the supply rails, IF your transformer is in spec and IF your AC mains is in spec. You can estimate the output power delivered to the loudspeakers, when the supply rails are ±28 volts DC, by studying Figure 37 of the LM3886 datasheet. You can also estimate the maximum heat dissipated by the LM3886 chip (and pulled away by an enormous heatsink conservatively chosen by YOU), when the supply rails are ±28 volts DC, by studying Figures 35 and 36 of the LM3886 datasheet.
_
Attachments
Yes, strangely, most of the thermal design considerations are described in other application notes like the AN-1192 Overture™ Series High Power Solutions.
Page 7 describes maximum supply voltages for bridge/parallel-cases for example.
Pages 3-6 are very interesting with regards to thermal data.
Also note, that if you have the isolated chip, its thermal resistance is a whopping 2K/W. At a theoretical Pdmax=40W that equates to +80K which is not only piping hot and hard to cool but a very unreliable application.
I´d guess many naive DIYers are driving those chips much too hard.
Thermal shutdown is at 150°C by the way! (page 4 of said app note)
Page 7 describes maximum supply voltages for bridge/parallel-cases for example.
Pages 3-6 are very interesting with regards to thermal data.
Also note, that if you have the isolated chip, its thermal resistance is a whopping 2K/W. At a theoretical Pdmax=40W that equates to +80K which is not only piping hot and hard to cool but a very unreliable application.
I´d guess many naive DIYers are driving those chips much too hard.
Thermal shutdown is at 150°C by the way! (page 4 of said app note)
I have built amps using LM3886 chips with a wide range of rail voltage. Some were unintentional. Bought a pair of -25,0, 25 volt 200 watt one time and had almost 38 volt rails. Chips ran hot. Switched to 24 volt Antek and got 31 volt rails. Perfect.
Rail optimal voltage will vary with speaker selection. For 8 ohm speakers 30 - 32 volt rails is fine. If 4 ohm speakers are used 28 - 30 volt will deliver more current.
Current amps use LM3886 chips with a fancy switch mode supply. A resonant mode supply. Works great. Assume a Meanwell would also.
Rail optimal voltage will vary with speaker selection. For 8 ohm speakers 30 - 32 volt rails is fine. If 4 ohm speakers are used 28 - 30 volt will deliver more current.
Current amps use LM3886 chips with a fancy switch mode supply. A resonant mode supply. Works great. Assume a Meanwell would also.
±40 V might be okay with an 8 Ω load, but ±30 is about the optimal with a 4 Ω load. At higher supply voltages the SPiKe SOA protection will engage earlier, which means that you get less output power into 4 Ω at voltages above ±30 V than you do at ±30 V.
My thoughts on LM3886 power supplies can be found here:
Taming the LM3886 Chip Amplifier: Output Power – Neurochrome
and in more detail here:
Taming the LM3886 Chip Amplifier: Power Supply Design – Neurochrome
Tom
It´s not a DV (design value") but an absolute maximum value.
Since it can´t be surpassed, and give parts tolerance, mains voltage variations,etc. ,safe design means you must keep *at least* a 10% safety margin.
And in any case, Datasheet clearly suggests +/-28V rails for 4 ohm and 8 ohm loads and +/-35V for 8 ohm and higher, NOT recommended for 4 ohm ones.
To boot, besides "standing" the supply voltage, chip must also dissipate associated thermal power, so stretching values up does not provide real advantage.
In any case, suggested power out is *excellent* for Home use, so why not enjoy it?
Thanks!
I know what is in the datasheet, but, I thought there are guys here experimented with different voltage levels for best sound.
I would have another question too: for the LM3886, do you guys think the speaker DC protection and delay circuit is needed? I would exclude it if it is very rare that LM3886 would die in a way that DC gets to the optput. If the swithc on and off knock is not huge, I do not care that either.
Thanks,
JG
I know what is in the datasheet, but, I thought there are guys here experimented with different voltage levels for best sound.
I would have another question too: for the LM3886, do you guys think the speaker DC protection and delay circuit is needed? I would exclude it if it is very rare that LM3886 would die in a way that DC gets to the optput. If the swithc on and off knock is not huge, I do not care that either.
Thanks,
JG
Voltage level does not change "sound", only maximum power output.
It is a standard amplifier, only in chipamp form for convenience.
It will fail in exactly same way as any other bipolar Class AB amplifier, including sending DC to speakers,so Turn on delay, DC protection, etc. can be a useful add on.
You decide.
It is a standard amplifier, only in chipamp form for convenience.
It will fail in exactly same way as any other bipolar Class AB amplifier, including sending DC to speakers,so Turn on delay, DC protection, etc. can be a useful add on.
You decide.
Thank You!
On the other hand, about voltage levels, it is pretty strange for me.
It was a long time ago when I built semiconductor based amps, for most of the time I had Mashall Leach's circuit based amps, with single and double transistor pair on the output too. I had both on +-30V and +-50V also and the difference was noticeable. The one with single pair output had better transient response with 50V, noticeable on percussion instruments for example, but not only on those. The one with double output I preferred with 30V.
I know this is "minor" and not measureable at all probably in distortion. On the other hand, in the gallery I see a several chipamp PCBAs built with Dale resistors. In my experience, that has much less of influence. Makes sense above a level, I agree, but at that level you would also care of the optimum rail voltages.
Also, on the protection, I see a lot of LM3886 based amps without. Of course it can fail in theory, but I'm interested about the experience. Has anyone had an LM3886 failed yet (without overstressing) ?
I actually have ready made DC protection / delay pcba, but if not needed, I would not complicate it, expecially I would not put the relay in the chain. I plan to use them for long term, I do not like to clean relay contacts over time. If, I use tranformes. I would use transformers but unfortunately my local source can not provide quality any more. That is why I considered switching PSU, but than, speaker protection will be needed. With a transformer it is pretty low chance that one side would fail, but the smpsu could shut down for some reason and if one side only, I'm not sure how the LM3886 reacts.
Thanks,
JG
On the other hand, about voltage levels, it is pretty strange for me.
It was a long time ago when I built semiconductor based amps, for most of the time I had Mashall Leach's circuit based amps, with single and double transistor pair on the output too. I had both on +-30V and +-50V also and the difference was noticeable. The one with single pair output had better transient response with 50V, noticeable on percussion instruments for example, but not only on those. The one with double output I preferred with 30V.
I know this is "minor" and not measureable at all probably in distortion. On the other hand, in the gallery I see a several chipamp PCBAs built with Dale resistors. In my experience, that has much less of influence. Makes sense above a level, I agree, but at that level you would also care of the optimum rail voltages.
Also, on the protection, I see a lot of LM3886 based amps without. Of course it can fail in theory, but I'm interested about the experience. Has anyone had an LM3886 failed yet (without overstressing) ?
I actually have ready made DC protection / delay pcba, but if not needed, I would not complicate it, expecially I would not put the relay in the chain. I plan to use them for long term, I do not like to clean relay contacts over time. If, I use tranformes. I would use transformers but unfortunately my local source can not provide quality any more. That is why I considered switching PSU, but than, speaker protection will be needed. With a transformer it is pretty low chance that one side would fail, but the smpsu could shut down for some reason and if one side only, I'm not sure how the LM3886 reacts.
Thanks,
JG
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.