I run a paralleled Lm4780 config (audiosector) which dispalys an odd behaviour. With no source connected I measure ~2-3V DC on both channels, however if I short the inputs (or connect an preamp) the DC offset is ~16mV on both channels. This is with the preamp off, with the preamp on, the only addition is some minor fluctuations when playing music. I`m using the amplifier and it works great, I was just confused why it would do this with no source? I use 0.1% resistors.
I haven't had speakers connected when I had the 2-3V DC, so no worries
I haven't had speakers connected when I had the 2-3V DC, so no worries
That schematic shows a DC coupled amplifier.
The output offset is ~ gain * ( input offset current * input resistance) + residual output offset of the amplifier.
For your arrangement the input resistance changes from Rin to Rsource when you couple up the pre-amp. That CHANGES the output offset.
If you cannot understand why you should not do that then you should not be building a DC coupled amplifier.
The output offset is ~ gain * ( input offset current * input resistance) + residual output offset of the amplifier.
For your arrangement the input resistance changes from Rin to Rsource when you couple up the pre-amp. That CHANGES the output offset.
If you cannot understand why you should not do that then you should not be building a DC coupled amplifier.
The schematic has a strange layout, which makes it a bit confusing. Cz is not part of the feedback loop.
If you mean to use Ci as per the datasheet, you should have a look into your sources. Most sources have an AC coupled output and in that case you can skip the blocking cap Cin and use only Ci. Since you have to adapt the corner frequencies of both input filters and all sources tend to have different filters, it is still better to use Cin and Ci. If you use DC coupled sources, you should not skip Cin, when you use Ci.
If you mean to use Ci as per the datasheet, you should have a look into your sources. Most sources have an AC coupled output and in that case you can skip the blocking cap Cin and use only Ci. Since you have to adapt the corner frequencies of both input filters and all sources tend to have different filters, it is still better to use Cin and Ci. If you use DC coupled sources, you should not skip Cin, when you use Ci.
Close.
The output offset is ~ gain * ( V+ - V-) + residual output offset of the amplifier.
Which is gain * (Zin+ * Iin+) - (Zin- *Iin-) where the I's are input bias current of the OPAMP and the Zins are the OPAMP input impedances .
So if the input Zs(which is the opamp and they are) and I's are matched at + and - in of the OPAMP you have no offset from this contribution.
The other factor is the resistances between grnd and the +/- inputs. The OPAMP input bias current * this resistance will give you a voltage drop, so if these resistances arent matched they will also cause an output offset. (FET inputs almost eliminate this problem)
In your cct. this last resistance changes from R7+R4 to R7+(R4//Rsource) Makes a diff, but I cant see it causing that much offset. Are you sure R4 is working (soldered properly, not defective)
Last edited:
Did you install the jumper: http://www.diyaudio.com/forums/audi...chip-amp-kits-dacs-chassis-13.html#post636571
I had my setup on the floor, just to see that everything worked and therefore I didnt have any ground connections from amplifier PCB:s, which I guess then makes R4 useless, so when I shorted the inputs the R4 got in the way and provided a load for the chips and therefore shrunk the DC offset to 16mV. So when I install it in a case I wont have this problem.
I didnt test my theory, but this seems plausibel?
Yes, the jumper is installed.
I didnt test my theory, but this seems plausibel?
Yes, the jumper is installed.
I have now installed the amplifier in a case and the problem persists, however, now only in one channel. With no source and without shorting the inputs I have 3.5V on one channel and 200mV on the other. If I short the inputs I get the same as before, which is 16mV (the same on both channels).
I`ve checked the input to ground resistor which is fine on both channels. I have cheched all the solders and I cant see any problems. I have a star ground with both amplifier PCB:s ground connections and the mains safety ground connected to it.
I`ve checked the input to ground resistor which is fine on both channels. I have cheched all the solders and I cant see any problems. I have a star ground with both amplifier PCB:s ground connections and the mains safety ground connected to it.
Tomten, the design has a certain properties and the main one is unbalanced impedances between the + input and the - input this together with rather high input bias currents.
If you draw (2 x input bias current x R4 x the gain) plus (input offset x gain) = can be quite much => a low impedance signal source (down to DC) is needed.
The current = 0.4-2 uA (from two chips)
Offset = 1-10 mV
If you have varying DC resistances you will get varying output offsets. Mr. Daniel has chosen to keep it simple and by that goes "simple" DC performance if you are unlucky. Tomten, if you check the AN-1192 and the datasheet for LM4780 you'll see that the engineers at National see the basic design differently.
If you draw (2 x input bias current x R4 x the gain) plus (input offset x gain) = can be quite much => a low impedance signal source (down to DC) is needed.
The current = 0.4-2 uA (from two chips)
Offset = 1-10 mV
If you have varying DC resistances you will get varying output offsets. Mr. Daniel has chosen to keep it simple and by that goes "simple" DC performance if you are unlucky. Tomten, if you check the AN-1192 and the datasheet for LM4780 you'll see that the engineers at National see the basic design differently.
Last edited:
peranders, I did your calculations and I see that the DC offset can be quite high but shouldnt the DC offset be a more reasonable value anyway, more like the other channel? Should it differ that much?
I have checked nationals LM4780 pdf and I can see that they use input capacitor and feedback capacitor, and I understand that this will affect DC offset (the feedback cap in this case), but isnt 3.5V a little of the scale even without them?
I have checked nationals LM4780 pdf and I can see that they use input capacitor and feedback capacitor, and I understand that this will affect DC offset (the feedback cap in this case), but isnt 3.5V a little of the scale even without them?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.