Also I mean that the bulb was pulsing with the amp during undervoltage protection.
If you mean to place 100 Ohm resistor on +IN pins in parallel by saying " terminate the input with a resistor" than yes!
To insert Ci (two caps for every amp in the chip) I need to disconnect PCB from amp and resolder components on PCB. If only one cap for all amps of the chip I can do it.
All resistors for chips are 0.1% tolerance.
However I don't understand what is a reason of close matching of resistors when I can change a gain setting resistance in whatever way (but not much), Ri and RB resistors are not equal or maybe a stability of resistors is in great importance here?
I've tried to power the amp once more and the DC offset on the chip where there was 134mV became 30mV On the other chip where was 31mV the DC offset start to rise from 10mV to 42 than (during 10 sec.) stopped. After that I heard some light cracking sound from the IC and seen some tiny smoke. After I discovered that I've made a mistake with polarities of big Cs caps on one of two chips. They are burned out. However this was on a second pair of chip. On the others the polarity is right. So when I checked polarities and resistors at first time they were ok. On second pair of chips I didn't.
May I ask you if big Cs caps are gone how I get know if small MKP 0.1uF caps which where placed on shorted (if polarity was mismatched) Cs caps are in good condition or broken?
Likewise could you tell me if I can listen to music without 2200uF Cs caps and with only 4x4700uF caps on PSU? There will be only a lack of bass or smth else? I'm not sure if it's allowable to test the bass without Cs caps.
Thanks.
If you mean to place 100 Ohm resistor on +IN pins in parallel by saying " terminate the input with a resistor" than yes!
To insert Ci (two caps for every amp in the chip) I need to disconnect PCB from amp and resolder components on PCB. If only one cap for all amps of the chip I can do it.
All resistors for chips are 0.1% tolerance.
However I don't understand what is a reason of close matching of resistors when I can change a gain setting resistance in whatever way (but not much), Ri and RB resistors are not equal or maybe a stability of resistors is in great importance here?
I've tried to power the amp once more and the DC offset on the chip where there was 134mV became 30mV On the other chip where was 31mV the DC offset start to rise from 10mV to 42 than (during 10 sec.) stopped. After that I heard some light cracking sound from the IC and seen some tiny smoke. After I discovered that I've made a mistake with polarities of big Cs caps on one of two chips. They are burned out. However this was on a second pair of chip. On the others the polarity is right. So when I checked polarities and resistors at first time they were ok. On second pair of chips I didn't.
May I ask you if big Cs caps are gone how I get know if small MKP 0.1uF caps which where placed on shorted (if polarity was mismatched) Cs caps are in good condition or broken?
Likewise could you tell me if I can listen to music without 2200uF Cs caps and with only 4x4700uF caps on PSU? There will be only a lack of bass or smth else? I'm not sure if it's allowable to test the bass without Cs caps.
Thanks.
If Ri and Rf are not the same on the parallel ICs, they will work on slightly different gain. One of the ICs will try to amplifiy more than the other, and at the outputs there will be constant battle for the right voltage. Voltage differences at the output will lead to a constant current flow from one IC to the other. They will heat up more than necessary. Therefore the resistors should be matched as tightly as possible to have the gain as equal as possible.AndriyOL said:
You can get into the same situation with perfectly matched gain resistors, if the DC offsets of the parallel ICs are too different. One of the reasons, why National uses DC servos in ther BPA-200 application.
That rise could be due to the small capacitor between the input pins being charged with the DC offset.AndriyOL said:
They should be okay, because the current that destroyed the electrolytics did not pass though them.AndriyOL said:
Yes, you can.AndriyOL said:
You will have less dynamic during pulses and there might be more noise on the rails. Nothing vital. It could sound weaker than with all capacitors connected.AndriyOL said:
Thanks for detailed explanation. I appreciate it.
Now I soldered 100uF on chip (the only which I had spare ones).
Even with replaced electrolytic bulb was light up. DC offset was on the contrary. At first chip output there was 29mV DC offset (with replaced 100uF), on the second chip DC offset was 66-70mV and that chip was warmer than previous. Strange this is. Maybe electrolytics cause the problem? Polarities of the last ones are ok. However, I'll try to replace them.
Could you tell me has the PSU consume any current without a load except toroid if it's ok? There are only caps in fact.
Now I soldered 100uF on chip (the only which I had spare ones).
Even with replaced electrolytic bulb was light up. DC offset was on the contrary. At first chip output there was 29mV DC offset (with replaced 100uF), on the second chip DC offset was 66-70mV and that chip was warmer than previous. Strange this is. Maybe electrolytics cause the problem? Polarities of the last ones are ok. However, I'll try to replace them.
Could you tell me has the PSU consume any current without a load except toroid if it's ok? There are only caps in fact.
AndriyOL said:
Electrolytics can cause changes in the DC offset. Offsets up to 100 mV are not unusual for Gainclones and their derivatives. They will not be the reason for your hum.
Could there be a ground loop somewhere? Are your input connectors isolated from the case?
Could the transformer be inducing hum? Some members with such an issue could cure it by turning the transformer or moving it around, others had to remove the transformers entirely from the amplifier case.
AndriyOL said:
Yes, the capacitors have a small consumption, because they are not ideal capacitors. Their ESR lets a very small current flow. It will be so small that it is hard to measure and relative to the transformer's consumption it is negligible. A big current will flow for a short time, when you switch the amplifier on or off, because the capacitors are charged or dicharged then.
Transformers under normal conditions with the amp are noiseless. Besides they are isolated from the amplifier with metal case. So EMI is not concern.
I didn't connect to chassis anything except ICs (without washers only compound) to heatsinks, but it is isolated from all other things even between each other. I have isolated one side of the chassis where heatsinks are mounted from other side by means of isolation washers. Mostly amp case is made from plastic. Input connectors are placed on plastic.
There is only one small ground loop. Signal ground of two chips is connected in parallel by one wire to two input connectors. Ground loop is within star ground and signal grounds of two chips.
May I place 100W 230VAC plus 40W 230VCA light bulbs in parallel and in series to toroid's primary to test the amp?
I didn't connect to chassis anything except ICs (without washers only compound) to heatsinks, but it is isolated from all other things even between each other. I have isolated one side of the chassis where heatsinks are mounted from other side by means of isolation washers. Mostly amp case is made from plastic. Input connectors are placed on plastic.
There is only one small ground loop. Signal ground of two chips is connected in parallel by one wire to two input connectors. Ground loop is within star ground and signal grounds of two chips.
May I place 100W 230VAC plus 40W 230VCA light bulbs in parallel and in series to toroid's primary to test the amp?
keep a selection of light bulbs for test purposes.
40W, 60W, 75W, 100W, 150W.
Start with the smallest to check the transformer wiring.
Probably use the same 40W bulb for checking the rectifier and the smoothing additions.
Try first 40W then 60W when you first connect the amplifier. Check all your voltages and prove that you have connected everything correctly.
If you have proved to yourself that all is OK, remove the test bulb and listen to the amp.
The higher the bias of the amp then the bigger bulbs come in handy when setting the bias, otherwise the bulb starts to glow and changes the line voltages too low for bias setting.
Most/all ClassAB amps can be set up with <=150W bulbs.
ClassA amps can be tested in ClassAB mode, but the final bias setting must be done without bulb protection.
40W, 60W, 75W, 100W, 150W.
Start with the smallest to check the transformer wiring.
Probably use the same 40W bulb for checking the rectifier and the smoothing additions.
Try first 40W then 60W when you first connect the amplifier. Check all your voltages and prove that you have connected everything correctly.
If you have proved to yourself that all is OK, remove the test bulb and listen to the amp.
The higher the bias of the amp then the bigger bulbs come in handy when setting the bias, otherwise the bulb starts to glow and changes the line voltages too low for bias setting.
Most/all ClassAB amps can be set up with <=150W bulbs.
ClassA amps can be tested in ClassAB mode, but the final bias setting must be done without bulb protection.
Hi, AndrewT
Today I've tried to power up all 4 LM4780 chips of the amp. Also I replaced again 2 electrolytic caps from 2.2mF to 100uF but results are the same. I have come to conclusion that all caps left are fine.
My measurements of DC offset of all ICs were as follows.
First pair of chips.
When testing with 40W bulb in series to toroid's primary the bulb was pulsing together with voltage. Under these circumstances it's was impossible to determine DC offset.
When testing with 100W bulb the bulb was light up and DC offset was: chip1 - 35mV (average value), chip2 - 128mV (average value).
When testing with 150W bulb the bulb was light up and DC offset was: chip1 - 32mV (average value), chip2 - 200mV (average value) and chip was warm.
Second pair of chips.
When testing with 40W bulb the pulse frequency of the bulb was slow like heartbeat.
When testing with 100W bulb the bulb was light up and speakers DC offset was: chip1 - 45mV (average value), chip2 - 570mV (average value). When testing with 100W bulb and nichrome wire the bulb was pulsing and pulse frequency was in two times faster than with 40W bulb.
When testing with 150W bulb DC offset was: chip1 - 41mV (average value), chip2 - 630mV (average value) and chip was warm enough. Nichrome wire on chip2 after some minutes of work was hot.
When I power up chips for some seconds toroids are humming, thus drawing much current.
When I used my amp in accordance with National parallel schematic with Ci (feedback) caps these two chips with the biggest DC offset values of every pair of chips described above had the biggest DC offset values from all others used with Ci caps and were 10.9mV and 12mV respectively.
Hence I guess these two chips are not good.
Could you tell me AndrewT am I right or mistake?
Btw, I don't know the meaning of "bias" term in electronics. Will you explain me?
According to datasheet of chip maximum quiescent power supply current is 170 mA. That's mean that for two chips 100W bulb must be enough not to glow it. But in my case it is not.First pair of chips. First pair of chips. Second pair of chips. Second pair of chips.
Today I've tried to power up all 4 LM4780 chips of the amp. Also I replaced again 2 electrolytic caps from 2.2mF to 100uF but results are the same. I have come to conclusion that all caps left are fine.
My measurements of DC offset of all ICs were as follows.
First pair of chips.
When testing with 40W bulb in series to toroid's primary the bulb was pulsing together with voltage. Under these circumstances it's was impossible to determine DC offset.
When testing with 100W bulb the bulb was light up and DC offset was: chip1 - 35mV (average value), chip2 - 128mV (average value).
When testing with 150W bulb the bulb was light up and DC offset was: chip1 - 32mV (average value), chip2 - 200mV (average value) and chip was warm.
Second pair of chips.
When testing with 40W bulb the pulse frequency of the bulb was slow like heartbeat.
When testing with 100W bulb the bulb was light up and speakers DC offset was: chip1 - 45mV (average value), chip2 - 570mV (average value). When testing with 100W bulb and nichrome wire the bulb was pulsing and pulse frequency was in two times faster than with 40W bulb.
When testing with 150W bulb DC offset was: chip1 - 41mV (average value), chip2 - 630mV (average value) and chip was warm enough. Nichrome wire on chip2 after some minutes of work was hot.
When I power up chips for some seconds toroids are humming, thus drawing much current.
When I used my amp in accordance with National parallel schematic with Ci (feedback) caps these two chips with the biggest DC offset values of every pair of chips described above had the biggest DC offset values from all others used with Ci caps and were 10.9mV and 12mV respectively.
Hence I guess these two chips are not good.
Could you tell me AndrewT am I right or mistake?
Btw, I don't know the meaning of "bias" term in electronics. Will you explain me?
AndrewT said:
According to datasheet of chip maximum quiescent power supply current is 170 mA. That's mean that for two chips 100W bulb must be enough not to glow it. But in my case it is not.First pair of chips. First pair of chips. Second pair of chips. Second pair of chips.
AndriyOL said:
Did you measure with an Ohm-meter that they are really isolated?
Did you check the channels one by one? And if so, did each of them alone lead to the same fault?
What about your self-wound transformer? Do you have any experience in transformer winding and calculation? Could there be something wrong with it, because the wires are too thin? Or you did not compensate for the short-circuit voltage? Or ... ?
Try to work systematically. First start up the power supply. Once that works, start up one amplifier channel. Leave the other disconnected. When that works as well, start with the other one.
If your amplifier worked according to the National schematic, return it to that state. Then make the changes towards the CarlosFM schematic one by one and see, what happens.
First add all those snubber capacitors to the power supply. Test. Then change the input filter. Test. Then change the feedback circuit. Test. Last add the 300 pF capacitor to the inputs. Test.
If you need more help
- Post a complete and correct schematic of what you really did, because nobody can follow your mix of CarlosFM and National PA-100 schematic and what got replaced with what, when and why.
- Post a complete and correct schematic of your grounding concept.
Pacific is right.
Test the assembly in stages and prove each stage is correct.
Remove everything and test just the transformer alone.
If the bulb glows something wrong.
If the bulb lights you have a serious fault. STOP and find the fault.
Chipamps have the bias set inside the chip but it varies a lot and is very low. Ignore this until you move beyond chipamps.
Test the assembly in stages and prove each stage is correct.
Remove everything and test just the transformer alone.
If the bulb glows something wrong.
If the bulb lights you have a serious fault. STOP and find the fault.
Chipamps have the bias set inside the chip but it varies a lot and is very low. Ignore this until you move beyond chipamps.
pacificblue said:
Yes I measured and they are isolated.
I didn't check channels one by one. I'll do it.
I got know how much winding the toroid need for 1VAC on secondary than calculated length of wire and did it.
When make windings on transformer core it is easy to power it up and measure the voltage on secondaries just to know if it need more windings or not. I used thick 2mm D enameled cooper wire.
I use transformer with 2A 250VAC fast blow fuse. When tested with bulbs I didn't.
I'll try to do what you recommend and draw schematics.
Thanks.
I have made a schematic of my amp guys. You can see it at http://profile.imageshack.us/user/andriyol/ and tell your conclusions, if you wish.
the Signal Ground looks OK.
The Audio Ground with all the decoupling taken back is no good.
Each pair of decoupling caps should be connected to each other. The smallest by the shortest route and the biggest by a necessarily longer route.
The junctions of all three pairs should be commoned, and then the common line should be taken to the Audio Ground. All these routes from Active device through decoupling to common to Audio Ground should be as short as possible.
The chipamp sees 100r//2k on it's inverting input.
and sees 330r+15k on it's non-inverting input.
This mixing of AC and DC coupling on the inputs is guaranteed to generate excess output offset.
the output offset will be ~={chipamp output offset +-[15200 * input offset current]}.
I bet the two chips have significantly different output offset and further that the difference in offset between the two chipamp will vary with operating conditions.
This is exactly what you don't need in a parallel set up.
Start by adding 1mF or 1.5mF cap to the 100r NFB leg before connecting to Signal Ground.
Next time keep it simple until you understand what your circuits are doing and what they need to perform to specification.
The Audio Ground with all the decoupling taken back is no good.
Each pair of decoupling caps should be connected to each other. The smallest by the shortest route and the biggest by a necessarily longer route.
The junctions of all three pairs should be commoned, and then the common line should be taken to the Audio Ground. All these routes from Active device through decoupling to common to Audio Ground should be as short as possible.
The chipamp sees 100r//2k on it's inverting input.
and sees 330r+15k on it's non-inverting input.
This mixing of AC and DC coupling on the inputs is guaranteed to generate excess output offset.
the output offset will be ~={chipamp output offset +-[15200 * input offset current]}.
I bet the two chips have significantly different output offset and further that the difference in offset between the two chipamp will vary with operating conditions.
This is exactly what you don't need in a parallel set up.
Start by adding 1mF or 1.5mF cap to the 100r NFB leg before connecting to Signal Ground.
Next time keep it simple until you understand what your circuits are doing and what they need to perform to specification.
Hi, Andrew.
I appreciate your reply. Maybe it's obvious from scientific view, but I'm not a specialist in this area. However I understand you partially unfortunately of what you advised to me.
If I'm not mistake you said that without feedback caps is no good. - But main goal of Carlos schematic is to use chipamp without FB caps as they cut off the sound significantly.
If you mean that lytic and poly FB caps' legs should be connected to each other than to Signal Ground, than in this concept I can't understand what you mean "The smallest by the shortest route and the biggest by a necessarily longer route".
If as three decoupling pairs you mean caps and snubbers on voltage rails from chip, National advises that all ground problems can be avoided by returning all grounds separately to a common point. I don’t have Audio Ground on my schematic. So, as Audio Ground I guess you mean Star Ground. If you remember when I registered at forum I had a low frequency response related amp problem with National parallel schematic and you was who tried to help. Than I used only one ground route from all PS decoupling caps of chip (for both voltage rails) to Star Ground. Now I suppose it was incorrect, considering mentioned above.
As “mixing of AC and DC coupling on the inputs” you mean that AC goes from Input to +IN and DC goes from ground to -IN, and propose to add FB caps! 1mF or 1.5mF will destroy high frequency response completely I suppose plus undesired noise. That is unacceptable for me. I like treble more than bass and chipamp can do it the best as I know.
Likewise I have a great desire to understand the term of "bias", its role and function in my amp. I'll be grateful if you enlighten me.
Thanks.
With best regards,
Andriy.
I appreciate your reply. Maybe it's obvious from scientific view, but I'm not a specialist in this area. However I understand you partially unfortunately of what you advised to me.
If I'm not mistake you said that without feedback caps is no good. - But main goal of Carlos schematic is to use chipamp without FB caps as they cut off the sound significantly.
If you mean that lytic and poly FB caps' legs should be connected to each other than to Signal Ground, than in this concept I can't understand what you mean "The smallest by the shortest route and the biggest by a necessarily longer route".
If as three decoupling pairs you mean caps and snubbers on voltage rails from chip, National advises that all ground problems can be avoided by returning all grounds separately to a common point. I don’t have Audio Ground on my schematic. So, as Audio Ground I guess you mean Star Ground. If you remember when I registered at forum I had a low frequency response related amp problem with National parallel schematic and you was who tried to help. Than I used only one ground route from all PS decoupling caps of chip (for both voltage rails) to Star Ground. Now I suppose it was incorrect, considering mentioned above.
As “mixing of AC and DC coupling on the inputs” you mean that AC goes from Input to +IN and DC goes from ground to -IN, and propose to add FB caps! 1mF or 1.5mF will destroy high frequency response completely I suppose plus undesired noise. That is unacceptable for me. I like treble more than bass and chipamp can do it the best as I know.
Likewise I have a great desire to understand the term of "bias", its role and function in my amp. I'll be grateful if you enlighten me.
Thanks.
With best regards,
Andriy.
Would you tell me what will happen when I add to my schematic 47uF as feedback caps and what the sound and distortion will be?
Thx.
Thx.
1/(2*PI*47µ*100) = ~34
You form a high-pass filter with a corner frequency of 34 Hz. That means you will severely bring your bass-performance down, because the filter will audibly reduce the level up to 5*34 = 170 Hz.
With such a small Ri you will need a much bigger capacitor. At least 330 µF or bigger. But then you also have to adapt the input filter...
You form a high-pass filter with a corner frequency of 34 Hz. That means you will severely bring your bass-performance down, because the filter will audibly reduce the level up to 5*34 = 170 Hz.
With such a small Ri you will need a much bigger capacitor. At least 330 µF or bigger. But then you also have to adapt the input filter...
Good explanation. I guess under this configuration it's not reasonable to place Ci.
Thanks pacificblue.
Thanks pacificblue.
Hi pacificblue.
I did what you recommend. PSUs work Ok. I powered them independently without any problems. I tested two amps separately which had 41mV and 630mv DC offset. With separated chips’ tests I got a following DC offsets without any input connected: for chip1 – 430mV (had before with cooperative use – 41mV) and chip2 – 530mV (had before with cooperative use – 630mV). When return to working National parallel schematic, the amp stops work when to disconnect Ci cap and direct Ri to ground. I’ve done it when had used National parallel schematic.
Regards.
I did what you recommend. PSUs work Ok. I powered them independently without any problems. I tested two amps separately which had 41mV and 630mv DC offset. With separated chips’ tests I got a following DC offsets without any input connected: for chip1 – 430mV (had before with cooperative use – 41mV) and chip2 – 530mV (had before with cooperative use – 630mV). When return to working National parallel schematic, the amp stops work when to disconnect Ci cap and direct Ri to ground. I’ve done it when had used National parallel schematic.
Regards.
Those values were taken with the CarlosFM configuration still? Did you check, if you meter also indicates AC at the outputs?
The 300 pF capacitors across the input pins (Cc1 and Cc2 in your schematic) were still there? Make a test without them.
Is DC at the input consistent with the output, i. e. 21 times smaller?
The 300 pF capacitors across the input pins (Cc1 and Cc2 in your schematic) were still there? Make a test without them.
Is DC at the input consistent with the output, i. e. 21 times smaller?
Then leave Ci in. Make it bigger, if you feel that it improves the sound. Increase Cin in the same relationship, i. e. if you double Ci, double Cin as well.
pacificblue said:
Yes, from carlosfm cofiguration!
pacificblue said:
You advise me to measure AC on outputs? I'll try.
pacificblue said:
300pF across inputs are still there. I'll try without them.
pacificblue said:
I don't know if DC at the input consistent with the output, because I didn't use or connect any input source still, only input caps are soldered.
pacificblue said:
You recommend to turn back to National schematic? Using Ci caps with this configuration won't be wise I guess.
- Status
- Not open for further replies.