Hm.. Now I don't get 35V on the output, but something is still wrong.
It starts at about 1000mV, goes slowly down to 0mV and continues down below the zero-line. When it turned to -100mV I pulled the plug. What is wrong now?
It starts at about 1000mV, goes slowly down to 0mV and continues down below the zero-line. When it turned to -100mV I pulled the plug. What is wrong now?
What time-scale is slow? Your description sounds like a process that depends on capacitor charging or discharging. Could be the smoothing caps charging up unsymmetrically or the muting capacitor charging up.
If there is no speaker connected, it should be safe to check how much further the voltage develops. If you had a short, there would either be no voltage at all or one of the rail voltages would be present from the start. Are you using a light-bulb tester or fuses to protect the amplifier from the worst?
If there is no speaker connected, it should be safe to check how much further the voltage develops. If you had a short, there would either be no voltage at all or one of the rail voltages would be present from the start. Are you using a light-bulb tester or fuses to protect the amplifier from the worst?
pacificblue said:
Maybe 10-15 sec from 1000mV to 0mV
Yes, a capacitator may be the problem. I don't know which or how to check them, though.
What would I get from checking this? Do you think it will set to a stable level after a while?
I have fuses!
One or several capacitor could have been damaged, when the first IC blew. Or it could simply be an effect of tolerances that lead to unsymmetric charging of the rails, hence a varying DC offset until both rails are charged to their maximum voltages. You can check with the other amplifer channel to see, if it shows a similar behaviour during the first 15 seconds after cycling the power.Ulew said:
Checking capacitors is difficult without adequate measuring equipment. You could replace them just to calm your conscience or you keep using them, until your amplifier's performance deteriorates notably.
If it stabilizes around 100 mV, you could choose to ignore it. While a solid state amp with such a DC offset would be regarded as rubbish, many chipamps work with DC offset values that even surpass 100 mV, and their owners are quite happy with the sound.Ulew said:
pacificblue said:
I'll check if it stabilizes at a good value, and if so I'll keep it that way. This is my first chipamp, so I'll consider it a test, and in the future I'll build a better one!
Offset after a long time?
Dear Ulew,
After a long time goes by...what offset remains? If I interpret National's data sheet correctly, it should be much closer to 10 mV that to 100 mV with the circuit you show, if everything is sane.
Some hints as to what may be going on...the 22 uF in the feedback loop may be left with a negative voltage during turn-off...it depends upon your power supply and the music applied during turn-off (and probably some other things that I'm ignoring...)
Anyway, negative voltage on the cap does 2 things...
1. hurts the cap
2. if it remains stored, then it would make vout positive at the next turn on until things stabilize.
Doug Self recommends a diode across the 22 uF feedback cap to prevent too much voltage from getting you in trouble...see Figure 33, D1 across C2, of this link:
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm#2
He (Mr. Self) would also recommend a much larger value of the cap to avoid low frequency distortion.
Dear Ulew,
After a long time goes by...what offset remains? If I interpret National's data sheet correctly, it should be much closer to 10 mV that to 100 mV with the circuit you show, if everything is sane.
Some hints as to what may be going on...the 22 uF in the feedback loop may be left with a negative voltage during turn-off...it depends upon your power supply and the music applied during turn-off (and probably some other things that I'm ignoring...)
Anyway, negative voltage on the cap does 2 things...
1. hurts the cap
2. if it remains stored, then it would make vout positive at the next turn on until things stabilize.
Doug Self recommends a diode across the 22 uF feedback cap to prevent too much voltage from getting you in trouble...see Figure 33, D1 across C2, of this link:
http://www.dself.dsl.pipex.com/ampins/dipa/dipa.htm#2
He (Mr. Self) would also recommend a much larger value of the cap to avoid low frequency distortion.
Re: Offset after a long time?
If you ensure that the input filter is set to F-3dB @ xHz, then the NFB lower leg capacitor must be set to F-3dB @ <=x/1.4 Hz.
This does two things that fit with Self's advice.
One: it minimises the AC voltage across the NFB cap.
Two: it gives the large value of capacitor to minimise the distortion that results from using an electrolytic in the NFB loop.
the input filter must define the passband of the power amplifier, not the NFB components.djoffe said:
If you ensure that the input filter is set to F-3dB @ xHz, then the NFB lower leg capacitor must be set to F-3dB @ <=x/1.4 Hz.
This does two things that fit with Self's advice.
One: it minimises the AC voltage across the NFB cap.
Two: it gives the large value of capacitor to minimise the distortion that results from using an electrolytic in the NFB loop.
Ulew, you have one working channel with 0V offset. If your other channel is built identically then it's behaviour should be close to that of the working one. Close, not identical, but maybe...
The chances are that you have some other error in the build of the faulty channel. It can be extraordinarily difficult to find such errors on occasion as they can be obscured by a temporary mental 'blindspot'. Some soldering problems also only reveal themselves after very close inspection or reflowing.
You have a working channel available to you. The first thing to do is to check again that the second channel is truly identical to the first. Check by measuring pin to pin continuity with a DVM and working around until you have covered every component's connection to every other one (the netlist).
By substituting components into the working channel, one by one, you can determine that the other components you have are the correct values and are working correctly. You just swap the components until the circuit stops working. Of course there are some risks, you may destroy some components while swapping them around and it is extremely tedious, but at the end of the process you will have 2 working sets of components and one working circuit, and the confidence to bully the other channel into working.
One way or another you can grind the problem out of existence.
Wakarimasu ka?
w
The chances are that you have some other error in the build of the faulty channel. It can be extraordinarily difficult to find such errors on occasion as they can be obscured by a temporary mental 'blindspot'. Some soldering problems also only reveal themselves after very close inspection or reflowing.
You have a working channel available to you. The first thing to do is to check again that the second channel is truly identical to the first. Check by measuring pin to pin continuity with a DVM and working around until you have covered every component's connection to every other one (the netlist).
By substituting components into the working channel, one by one, you can determine that the other components you have are the correct values and are working correctly. You just swap the components until the circuit stops working. Of course there are some risks, you may destroy some components while swapping them around and it is extremely tedious, but at the end of the process you will have 2 working sets of components and one working circuit, and the confidence to bully the other channel into working.
One way or another you can grind the problem out of existence.
Wakarimasu ka?
w
- Status
- Not open for further replies.