Alright then. I measured the resistance of the 50K pots, and adjusted them all to 25K according to my multimeter. From then i measured the voltage across pin 1 and pin 3, got 36.2VDC across all the LM3886 chips.
I measured the VDC across pin 3 and (ground) on all the chips, got between +30mV and +40mV offset. Now i dunno if it matters, i'm measuring +16mV across the chips now, but in coming this far i've reached the extreme of the pot adjustment. I'm measuring a resistance of 51K ohm on the pots now, can go no further, leaving the DC offset to 16mV. I'm running them at the moment, no problems. But i'll do a second measurment tomorrow to concur with what i have.
rick
I measured the VDC across pin 3 and (ground) on all the chips, got between +30mV and +40mV offset. Now i dunno if it matters, i'm measuring +16mV across the chips now, but in coming this far i've reached the extreme of the pot adjustment. I'm measuring a resistance of 51K ohm on the pots now, can go no further, leaving the DC offset to 16mV. I'm running them at the moment, no problems. But i'll do a second measurment tomorrow to concur with what i have.
rick
An AC coupled single chip amp can be set up for near zero output offset. From cold at switch on (power up) to warm to hot the output offset will vary. The further away from zero offset, the worse the drift with temperature.
A DC coupled chipamp needs extra help to avoid risking damage. Don't use DC coupled nor mixed AC & DC coupled until you have learned how to cope with excessive output offset.
In the meantime, the 3 parallel chipamp is overall an AC coupled chipamp. But inside the AC coupling the chip are DC connected. There is the problem.
The output offset of each chip amp must, as closely as possible, match it's partners not only at start up but throughout all it's operating temperatures.
I suggest since you have 8ohm speakers that you experiment with a single AC coupled chipamp and find out what you can do to adjust output offset, how to adjust to zero offset and examine how much drift in offset is typical of a well set up chipamp.
When you are familiar with what to do and what to expect, then you are ready to look at adjusting your parallel chipamp amplifier.
If you do it wrong now you will either damage the amplifier or worse damage the speaker. Hell there is worse, you could damage both !
WARNING
paralleled (PA100) and bridged (BA100) and bridged paralleled (BPA300) chipamps are not for beginners. Never ever, unless they are guaranteed to survive beginner use.
A DC coupled chipamp needs extra help to avoid risking damage. Don't use DC coupled nor mixed AC & DC coupled until you have learned how to cope with excessive output offset.
In the meantime, the 3 parallel chipamp is overall an AC coupled chipamp. But inside the AC coupling the chip are DC connected. There is the problem.
The output offset of each chip amp must, as closely as possible, match it's partners not only at start up but throughout all it's operating temperatures.
I suggest since you have 8ohm speakers that you experiment with a single AC coupled chipamp and find out what you can do to adjust output offset, how to adjust to zero offset and examine how much drift in offset is typical of a well set up chipamp.
When you are familiar with what to do and what to expect, then you are ready to look at adjusting your parallel chipamp amplifier.
If you do it wrong now you will either damage the amplifier or worse damage the speaker. Hell there is worse, you could damage both !
WARNING
paralleled (PA100) and bridged (BA100) and bridged paralleled (BPA300) chipamps are not for beginners. Never ever, unless they are guaranteed to survive beginner use.
Last edited:
well i believe i have come as far as i can for this exercise and this chip amp. The lowerst offset i'm able to achive is 16mV -/+0.5mv. I'm finding trouble when i adjust one pot, it throws one of the other two out. I've come this far and have a 16mV offset with a -/+0.5mV variance between the three chips. I dunno, how am i doing? :\
Last edited:
then when you finally connect a load, make it a cheap dummy load.
8off ohm 600mW resistors would do just fine. Monitor what happens to them over an extended test period, through a variety of different ON to OFF periods, cold and warm and hot start ups, overnight running to experience a bigger range of mains voltages.
Find out how it performs before you connect any speaker and certainly before you connect expensive speakers.
8off ohm 600mW resistors would do just fine. Monitor what happens to them over an extended test period, through a variety of different ON to OFF periods, cold and warm and hot start ups, overnight running to experience a bigger range of mains voltages.
Find out how it performs before you connect any speaker and certainly before you connect expensive speakers.
Hi Richard,
I'm a surprized at the results you're getting. Did you ever receive a reply from the seller for a schematic? Without that we're just shooting in the dark. In a well designed circuit you shouldn't need to have the trimmer all the way to the end stop to just get it close. It might be a good idea at this point to remove two of the three amp chips and see if it's possible to adjust the offset correctly on just one of them, it seems you're getting a lot more interaction than would be reasonable. And I agree with Andrew on this, until you are totally confident it's working right, you shouldn't hook it to expensive speakers, use a dummy load or some cheap speaker you're not afraid to sacrifice.
Mike
I'm a surprized at the results you're getting. Did you ever receive a reply from the seller for a schematic? Without that we're just shooting in the dark. In a well designed circuit you shouldn't need to have the trimmer all the way to the end stop to just get it close. It might be a good idea at this point to remove two of the three amp chips and see if it's possible to adjust the offset correctly on just one of them, it seems you're getting a lot more interaction than would be reasonable. And I agree with Andrew on this, until you are totally confident it's working right, you shouldn't hook it to expensive speakers, use a dummy load or some cheap speaker you're not afraid to sacrifice.
Mike
Hi Mike, still waiting for them to send the schematic, also maybe the muiltimeter is giving wrong reading, i'll get hold of another today to make reference. And yes i too thought of removing one or two of the chips to see if i could get it down further. You stated it was not needed to have three of them, if removing one makes my situation easier i will remove one. Could this problem have something to do with my floundering around previously with the power supply?
(off to work back in 8hrs)
rick
(off to work back in 8hrs)
rick
I recommend removing two of them and leaving just one, for driving an 8 ohm load that's all you need anyway. That way you will eliminate any potential for interaction between chips, which will make it much easier to evaluate circuit performance. I don't know if anything was damaged or not, but these chips tend to be fairly rugged and not easily harmed. I'm also concerned about fake or relabeled chips, it has been an issue, especially when they are bought so cheaply. 
Mike
Mike
Last edited:
update on this. I've removed one of the chips (middle one) from the circuit, this helped improve the DC offset so far, but unable to reach 0mV on both chips, it's possible to reach 0mV with one but i throws the other one off target still, but not as bad as it was with three chips. Two chips are much more stable to set than three, and i've managed to get an offset of 4.5mV on all the remain 4 chips.
And what i'm thinking instead of removing the last chip i may replace the 50K pot's with a 100K pots, or try to add some extra 25Kohms to the circuit along with with the 50K pot in place and try again from there. It seems that the more resistance added with the pot, the more balanced and tighter the two chips measure to zero. I'm getting there.............
And what i'm thinking instead of removing the last chip i may replace the 50K pot's with a 100K pots, or try to add some extra 25Kohms to the circuit along with with the 50K pot in place and try again from there. It seems that the more resistance added with the pot, the more balanced and tighter the two chips measure to zero. I'm getting there.............
Last edited:
That's one of the important lessons that I was referring to.
I would prefer the added 20k to 30k rather than a higher value VR.
Recognising what affects offset and being able to formulate a plan to attenuate the drift in offset is very important to any amp. But not knowing that in a paralleled system risks damage.
Now I am going into an area I have not experimented with: At what temperature do you ZERO the offset of the paralleled chipamp.
Do you zero the offset of all chips when completely cold? i.e. at ambient temperature (which will vary from season to season).
This will minimise the imbalance in startup currents. and allow the chips to start without excessive interchip current flow.
or
Do you zero the offset when fully warmed up (or even hot while playing music)? This minimises the imbalance while operating. But will the chips survive a cold start up?
If I were doing this (and I never will parallel chipamps) I would cold zero the offset and then use DC servo on every chip to maintain that good balance throughout the warming up period. This is quite a bit more complicated than the three VRs supplied, but is the National solution.
Last edited:
If the VDC input were to be reduced could that inturn lower the DC offset i'm measuring. The circuit boards i have, have +30VDC -30VDC printed on them, i'm inputing 36VDC each rail, could that extra 6VDC be pushing the circuit and them 50K pots out of range a little. And attached is the schematic sent by the company that produces the PCB for the amp. Is the pot represented by R4, i'm sure it is. replacing that could be a minor task.
Attachments
No, the output offset is determined by the internal architechture of the chipamp and a few external component values. Voltage will not have a significant effect. A bootstrap driven VAS is very susceptible to supply voltage variations. I doubt any chip designer uses that topology.
Input offset current, device temperature and -IN & +IN source resistances determine the most significant variations in offset.
The 50k does not have sufficient range to take account of the very variable parameter, input offset current, that exists in cheaply manufactured (untrimmed) devices.
Some very expensive and a few not so expensive opamps have this trimming done at the chip level.
Input offset current, device temperature and -IN & +IN source resistances determine the most significant variations in offset.
The 50k does not have sufficient range to take account of the very variable parameter, input offset current, that exists in cheaply manufactured (untrimmed) devices.
Some very expensive and a few not so expensive opamps have this trimming done at the chip level.
So in saying that, would it be safe (other than removing another chip) to replace those pots with one of a larger range, and go from there.
Far cheaper to insert an extra resistor. And if you have guessed wrong to change that resistor.
You already know that 50k is too low for some of your chipamps. That tells me and should tell you that the largest fixed resistor you can fit will be 47k.
If that turns out to leave the 50kVR very near the zero resistance end, you can easily add another resistor (maybe a second 47k) in parallel to move the usable region for the wiper to nearer the middle, leaving lots of leeway for further adjustment.
You already know that 50k is too low for some of your chipamps. That tells me and should tell you that the largest fixed resistor you can fit will be 47k.
If that turns out to leave the 50kVR very near the zero resistance end, you can easily add another resistor (maybe a second 47k) in parallel to move the usable region for the wiper to nearer the middle, leaving lots of leeway for further adjustment.
Wow, what a mess that design is, in my opinion it's a disaster waiting to happen. AndrewT is right, you need more resistance range to work with, but because the board is small and uses surface mount resistors, it would probably be a lot easier to just replace the 50k trim pots with 100k ones, otherwise you'll have to modify the board which could get messy. Whatever you end up doing, I urge you to use a dummy load or a cheap speaker to test it thoroughly before putting it to use with expensive speakers. I also recommend taking a look at this, parallel amp sectoin starting on page seven: http://www.national.com/an/AN/AN-1192.pdf
Mike
Mike
thanks mate. One silly question; What is it i'll be testing for? when i do test them.
I had them running on a 4.7ohm resistor tonight for a few hours, tested the DC offset a few times during and after, and it remained the same through out.
And i will be changing the pots as well, i believe that will help alot the achive a better DC offset.
I had them running on a 4.7ohm resistor tonight for a few hours, tested the DC offset a few times during and after, and it remained the same through out.
And i will be changing the pots as well, i believe that will help alot the achive a better DC offset.
Sounds like you're on the right track, checking for offset. Only thing I would add for testing is to run them through several temperature cycles, check them cold right after power up, then check after fully warmed up, power down and allow to cool. Repeat that pattern several times and keep track of you measurements and see if it's consistant. If there isn't a lot of variation, it should be OK to go ahead and use them.
Mike
Mike
Alright second round of testing. And i'm measuring 4.0~4.5mVDC from the chips after a couple hours of operation. Though when i measure on cold start up i measure some ~5mV difference between the chips, that does pull back to a ~0.5mV difference after a few minutes of warm up. And tempuratures from the heatsinks are between 42C and 44C consistantly. So i'm doing alright here.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.