It is important that is has a low DC impedance or the noise goes up. Q and self resonance of such small coils is not an issue. It is in the several MHz range.
I do not think so, but you can try. RF is a strange thing. A friend of mine is a master of ferrite tubes and pills. I use them only when there is an actual problem.
I thinck the only thin that can oscillate here is the buffer...It's very very wide-band...so it may be good idea to build a R-C filter..there or bypass the neuman resistor with a small cap that rolls off the signal above a few hundred KHz..
Hesener
How did you notice the oscilations this time ? You said these become aparent when using long cables, but where you listening to music ?
Are these audible ?
How did you notice the oscilations this time ? You said these become aparent when using long cables, but where you listening to music ?
Are these audible ?
Hi Joachim,
tonight I had the mpp2011 boards on the bench. One board, the red painted transistor one, seems to work ok (after adjusting R4 to equal the input leg sources), the cascode ref Voltages are around 6V and waveforms are clean (riaa typical, resp.) (did not check output offset voltage on that board, grmpf)
The other board had an issue with the shunt cascode voltage reg in the negative voltage twig, fixed that (had -1.3V instead of -6V at the cascodes base trannie), but this board has more issues: before the output buffer, there is -7V offset, and at the output -15V. No clean waveforms to get. There must be both in the output buffer and something in that nasty main current mirror going on. Or the previous imbalance has destroyed something. I did not manage to measure a meaningful current in the output buffer fets, (it wandered around 7ma per twig and quad, a bit lean, I'd say)
Rüdiger
tonight I had the mpp2011 boards on the bench. One board, the red painted transistor one, seems to work ok (after adjusting R4 to equal the input leg sources), the cascode ref Voltages are around 6V and waveforms are clean (riaa typical, resp.) (did not check output offset voltage on that board, grmpf)
The other board had an issue with the shunt cascode voltage reg in the negative voltage twig, fixed that (had -1.3V instead of -6V at the cascodes base trannie), but this board has more issues: before the output buffer, there is -7V offset, and at the output -15V. No clean waveforms to get. There must be both in the output buffer and something in that nasty main current mirror going on. Or the previous imbalance has destroyed something. I did not manage to measure a meaningful current in the output buffer fets, (it wandered around 7ma per twig and quad, a bit lean, I'd say)
Rüdiger
Last edited:
Hi Rüdiger !
Sam told me that he got some transistors from e-bay that may be fakes. It may also be that Sam did not select Idss well or simply put in the wrong group. I know, this is a frustrating repair so kudos to your brans.
Anyway, one channel works so there is no mistake in layout or topology.
The BJts can be checked with a diode tester for Ube. That usually shows if they work or not. One way around you should measure something like 0,65V and the other way around it should not conduct, provided there is no resistor bypass or loaded cap. PNP are of cause conducting in the opposite position then the NPNs. I just mention that because i am intermingling this issue sometimes too. To test the Fets ? Solder them out piece by piece.
Maybe you can compare voltages to channel that works, also from point to point.
Sam put in some extremely low ESR polymer elcaps. They have a huge leakage current that can be problematic. I substituted with Elna Silmik 2. Did i do that in both channels ?
Sam told me that he got some transistors from e-bay that may be fakes. It may also be that Sam did not select Idss well or simply put in the wrong group. I know, this is a frustrating repair so kudos to your brans.
Anyway, one channel works so there is no mistake in layout or topology.
The BJts can be checked with a diode tester for Ube. That usually shows if they work or not. One way around you should measure something like 0,65V and the other way around it should not conduct, provided there is no resistor bypass or loaded cap. PNP are of cause conducting in the opposite position then the NPNs. I just mention that because i am intermingling this issue sometimes too. To test the Fets ? Solder them out piece by piece.
Maybe you can compare voltages to channel that works, also from point to point.
Sam put in some extremely low ESR polymer elcaps. They have a huge leakage current that can be problematic. I substituted with Elna Silmik 2. Did i do that in both channels ?
Ricardo, if it is the buffer you should jumper the Neumann resistor ( you do that anyway ).
Then you have to insert a 220 Ohm resistor where it goes out of the RIAA into the bases of the buffer and a 150pF cap from the bases to ground. That limits the buffer to ca. 2Mhz -3dB. The buffer itself is so incredible fast that is passes a 1MHz square.
Then you have to insert a 220 Ohm resistor where it goes out of the RIAA into the bases of the buffer and a 150pF cap from the bases to ground. That limits the buffer to ca. 2Mhz -3dB. The buffer itself is so incredible fast that is passes a 1MHz square.
Hi,
no, elnas are only in one channel, the other has computer mainboard looking ones...
I usually do the Ube testing with my multimeter. Occasionally, I short b and e and transistors tend to dislike this, if some current comes into play....
On the not working board some input fets have obviously previously been replaced. I'm getting a bit scroogy with my last fets...
Rüdiger
no, elnas are only in one channel, the other has computer mainboard looking ones...
I usually do the Ube testing with my multimeter. Occasionally, I short b and e and transistors tend to dislike this, if some current comes into play....
On the not working board some input fets have obviously previously been replaced. I'm getting a bit scroogy with my last fets...
Rüdiger
I can send you the Fet´s you need, also P-Channel. Send me a mail. Sam should have some too. His Avatar here is "Masag". Yes, BJTs can start to avalanche when you do that and then they get noisy. That is the reason i double check each semiconductor before i stuff the board. Patience is saving time in this occasion. Even with precaution i managed to miss stuff some transistors in the Paradise i build.
Ricardo, there is another way to compensate the buffer that does not require to cut tracks and insert a resistor. Simply bypass the buffer from input bases to output after the 15 Ohm resistors,. That bootstrap reduces the current gain to zero at high frequencies but keeps the voltage gain that is aprox. 1x. A 15pF mica or COG should do. When you really want to optimize the impulse response of the buffer in the timedomain you have to isolate the buffer by not stuffing the RIAA and not connecting the outputs of the mirrors to the bases of the buffer. You than need a say 47kOhm resistor from bases buffer to ground to establish a DC current path and use the input of the buffer as input of your generator. Put in a 100kHz square at ca 1V and measure the square wave with a scope at the output. Adjust the bootstrap cap for minimum overshot of the square. Sometimes you need the input RC filter too or you have to insert a low value resistor in series with the bootstrap cap. In Praxis, with the RIAA in place the signal is shunted to ground anyway by the "treble" cap so i do not expect a major oszillation problem of the buffer. What can happen is that the PCB tracks from RIAA to input of the buffer, although short can produce a resonance with the input impedance of the buffer. That is damped of cause by the "bass" resistor and so it goes.....
Ricardo, there is another way to compensate the buffer that does not require to cut tracks and insert a resistor. Simply bypass the buffer from input bases to output after the 15 Ohm resistors,. That bootstrap reduces the current gain to zero at high frequencies but keeps the voltage gain that is aprox. 1x. A 15pF mica or COG should do. When you really want to optimize the impulse response of the buffer in the timedomain you have to isolate the buffer by not stuffing the RIAA and not connecting the outputs of the mirrors to the bases of the buffer. You than need a say 47kOhm resistor from bases buffer to ground to establish a DC current path and use the input of the buffer as input of your generator. Put in a 100kHz square at ca 1V and measure the square wave with a scope at the output. Adjust the bootstrap cap for minimum overshot of the square. Sometimes you need the input RC filter too or you have to insert a low value resistor in series with the bootstrap cap. In Praxis, with the RIAA in place the signal is shunted to ground anyway by the "treble" cap so i do not expect a major oszillation problem of the buffer. What can happen is that the PCB tracks from RIAA to input of the buffer, although short can produce a resonance with the input impedance of the buffer. That is damped of cause by the "bass" resistor and so it goes.....
Rüdiger, the "Computer" caps are the polymers.
It is not necessary to use Elnas. Panasonic FC wil do.
It is not necessary to use Elnas. Panasonic FC wil do.
I'll check if any Q's are faulty. It will take time, though...
I'll send mail if a certain quantity of fets should be dead.
Rüdiger
I'll send mail if a certain quantity of fets should be dead.
Rüdiger
I have a audio test box (basically, a sound card with extra connectors), where I have a scope running in parallel connected to its output, and when I connect the cable to the input of Paradise I can see the scope go wild... it didnt happen when listening though. So, I think it might be due to large input capacitance, but in the moment it happened I didnt consider trying to just connect the scope and cable, to see if that causes oscillation... my fault... will do that on friday
That is very true. I do have the load impedance in parallel to the input as well, and for my measurements that was set to 1000 Ohm. Will investigate further....